Atlasafricanag All 1

ATLAS
OF AFRICAN
AGRICULTURE
RESEARCH &
DEVELOPMENT
ATLAS
OF AFRICAN
AGRICULTURE
RESEARCH &
DEVELOPMENT
Edited by Kate Sebastian
A peer-reviewed publication
International Food Policy Research Institute
Washington, DC
ATLAS
OF AFRICAN
AGRICULTURE
RESEARCH &
DEVELOPMENT
ABOUT THE MAPS

Where not otherwise cited, the administrative boundaries and names shown and the designations
used on the maps are from Global Administrative Unit Layers 2013 from the Food and Agriculture
Organization of the United Nations (www.fao.org/geonetwork/). The use of these data does not imply
official endorsement or acceptance by the International Food Policy Research Institute; the CGIAR
Consortium; the CGIAR Research Program on Climate Change, Agriculture and Food Security; the
Food and Agriculture Organization of the United Nations; the Bill & Melinda Gates Foundation; or any
of the contributing authors or institutions.
The 2013 boundaries and names used on the maps for Ruminant Livestock and Map 2 of Statistical
Groupings were provided by the World Banks Map Design Unit.
Where not otherwise cited, a total population figure for Africa of 1.03 billion was used, based
on the United Nationss estimate for 2010 from The World Population Prospects: The 2012 Revision
(http://esa.un.org/wpp/Excel-Data/population.htm). For mapping and analysis, the Global Population
of the World (GPW) population data projected for 2010 from the Center for International Earth
Science Information Network and the Centro Internacional de Agricultura Tropical (http://sedac.ciesin.
columbia.edu/data/set/gpw-v3-centroids, accessed Feb. 4, 2014) were used.
Any opinions stated herein are those of the authors and are not necessarily representative of or
endorsed by the International Food Policy Research Institute or any of the partner organizations.
Copyright 2014 International Food Policy Research Institute. All rights reserved.
Contact ifpri-copyright@cgiar.org for permission to reprint.
2033 K Street, NW
Washington, DC 20006-1002, USA
Telephone: +1-202-862-5600
www.ifpri.org
DOI: http://dx.doi.org/10.2499/9780896298460
Library of Congress Cataloging-in-Publication Data

Atlas of African agriculture research and development / edited by Kate Sebastian.
pages cm
Includes bibliographical references.
ISBN 978-0-89629-846-0 (alk. paper)
1. Agriculture--Economic aspects--Africa. 2. Agriculture--Research--Africa. 3. Geospatial data-Africa. I. Sebastian, Katherine L. II. International Food Policy Research Institute.
HD2118.A87 2014
338.1096--dc23
2014008351
COVER PHOTO CREDITS

Photos (l): Globe: Hemera-Thinkstock; and starfield: Photodisc, C. Banker/J. Reed/Thinkstock.
Photos (r, top-bottom): Maize in South Africa: Panos/J. Larkin; cassava field: Media Bakery/S. Paname/
Veer; village women and livestock in Niger: ILRI/S. Mann; road market in Tanzania: Media Bakery/A.
Shalamov/Veer; womens farming cooperative in the Democratic Republic of the Congo: Panos/G.
Pirozzi; grain for food, straw for feed: ILRI/Gerard; carrying bags of coffee in Uganda: Panos/S. Torfinn.
Cover design: Anne C. Kerns, Anne Likes Red, Inc.
Book design and layout: David Popham, IFPRI
Editor: Sandra Yin, IFPRI
Contents
Foreword vii
Acknowledgments ix
Abbreviations and Acronyms xi
Introduction xiii
Political, Demographic, and Institutional Classifications 1
Administrative Boundaries 2
Statistical Groupings 4
Public Agriculture R&D Investments 6
Africas Agricultural Research Pool 8
CGIAR Research Program on Dryland Systems 10
Works Cited 12
Footprint of Agriculture 13
Farming Systems of Africa 14
Cropland and Pastureland 16
Irrigated Areas 18
Cereal Crops 20
Root Crops 22
Livestock and Mixed Crop-Livestock Systems 24
Ruminant Livestock 26
Cropping Intensity 28
Land Productivity for Staple Food Crops 30
Works Cited 32
Growing Conditions 33
Agroecological Zones 34
Climate Zones for Crop Management 36
Rainfall and Rainfall Variability 38
Soil Fertility 40
Works Cited 42
Role of Water 43
Effects of Rainfall Variability on Maize Yields 44
Blue and Green Virtual Water Flows 46
Blue and Green Water Use by Irrigated Crops 48
Rainfall Data Comparison 50
Works Cited 52
Drivers of Change 53
Influence of Aridity on Vegetation 54
Impacts of Climate Change on Length of Growing Period 56
Maize Yield Potential 58
Wheat Stem Rust Vulnerability 60
Benefits of Trypanosomosis Control in the Horn of Africa 62
Works Cited 64
Access to Trade 65
Market Access 66
Accessing Local Markets: Marketsheds 68
Accessing International Markets: Ports and Portsheds 70
Works Cited 72
Human Welfare 73
Severity of Hunger 74
Poverty 76
Early Childhood Nutrition and Health 78
Works Cited 80
About the Authors 81

Glossary 87
vi
Foreword
Africa is a paradox. This vast continent is home to almost half of the worlds uncultivated
land fit for growing food cropsan estimated 202 million hectaresbut much of it is off
limits to farmers because it is difficult to farm or it is used for other purposes. Despite some
recent economic successes, nearly a quarter of its population suffers from hunger, and Africa
has the highest incidence of poverty in the world.
It has long been recognized that Africa needs to significantly and sustainably intensify its
smallholder agriculture. Low-input, low-productivity farming has failed to keep pace with
food demands from a rising population. But achieving sustainable increases in smallholders
productivity is not easy. In many areas erratic rainfall, poor soil fertility, and a lack of infrastructure and support services offer limited prospects and few incentives for poor farmers to
invest in boosting productivity.
Comparing and contrasting where the challenges to and opportunities for growth in productivity are located, and doing so at multiple scales and over time, can give us powerful insights
that can enrich our understanding of the variables that affect agricultural productivity. The
Atlas of African Agriculture Research & Development presents a broad range of geospatial data
that relate to strategic agriculture policy, investment, and planning issues. The maps and
analyses will give anyone who wants to learn about the role of agriculture in Africa, or find
new ways to boost agricultural performance, a sense of the increasingly diverse geospatial
data resources that can inform their work and guide decisionmaking on agricultural development. A better understanding of current and evolving growing conditions and how to
increase productivity, despite obstacles, should aid in tailoring more pragmatic solutions for
poor smallholder farmers.
Shenggen Fan
Director General
vii
Acknowledgments
This atlas was supported by funding from the Bill & Melinda Gates Foundation (BMGF),
HarvestChoice, the CGIAR Consortium for Spatial Information (CGIAR-CSI), and the CGIAR
Research Program on Climate Change, Agriculture and Food Security (CCAFS).
An atlas covering such a broad set of topical issues on agricultural research and development in Africa required the expertise of many people who generously offered their time and
insights. The editor, Kate Sebastian, and publisher, the International Food Policy Research
Institute (IFPRI), wish to thank the following participating authors for their contributions
and work as we created, refined, and finalized the content of the atlas: Christopher Auricht
at Auricht Properties, Carlo Azzarri at HarvestChoice/IFPRI, Jason Beddow at the University
of Minnesota (UMN), Nienke Beintema at Agricultural Science and Technology Indicators
(ASTI)/IFPRI, Chandrashekhar Biradar at the International Center for Agricultural Research
in Dry Areas (ICARDA), Jean-Marc Boffa at the World Agroforestry Centre (ICRAF), Giullano
Cecchi at the Food and Agriculture Organization of the United Nations (FAO), Yuan Chai
at UMN, Guiseppina Cinardi at FAO, Lieven Claessens at the International Crops Research
Institute for the Semi-Arid Tropics (ICRISAT), Guilia Conchedda at FAO, Cindy Cox at
HarvestChoice/IFPRI, John Dixon at the Australian Centre for International Agricultural
Research (ACIAR), Petra Dll at Goethe University, Kathleen Flaherty at ASTI/IFPRI, Karen
Frenken at FAO, Heidi Fritschel at IFPRI, Dennis Garrity at ICRAF, Marius Gilbert at Universit
Libre de Bruxelles, Zhe Guo at HarvestChoice/IFPRI, Jawoo Koo at HarvestChoice/IFPRI,
Raffaele Mattioli at FAO, Tolulope Olofinbiyi at IFPRI, Felix Portmann at Senckenberg
Research Institute, Navin Ramankutty at McGill University, Timothy Robinson at the
International Livestock Research Institute (ILRI), Alexandra Shaw (consultant), Stefan Siebert
at the University of Bonn, Gert-Jan Stads at ASTI/IFPRI, Philip Thornton at CCAFS/ILRI,
Antonio Trabucco at Euro-Mediterranean Center on Climate Change (CCMC), Justin Van
Wart at the University of Nebraska, Klaus von Grebmer at IFPRI, Doris Wiesmann (consultant), William Wint at the Environmental Research Group Oxford, Stanley Wood at BMGF,
Ulrike Wood-Sichra at HarvestChoice/IFPRI, Sandra Yin at IFPRI, Yisehac Yohannes at IFPRI,
and Robert Zomer at ICRAF-China.
Although each map theme is attributed to the contributing authors and their respective
organizations, the authors would like to acknowledge the following for helping make this
atlas possible:
Terrance Hurley and Philip Pardey at UMN, and Darren Kriticos at Commonwealth
Scientific and Industrial Research Organisation (CSIRO), who contributed to Wheat
Stem Rust Vunerability;
Jusper Kiplimo and An Notenbaert at ILRI, who assisted with mapping and analysis
related to Livestock and Mixed Crop-Livestock Systems, Rainfall and Rainfall
Variability, and Impacts of Climate Change on Length of Growing Period;
Peter Jones at Waen Associates, for his input on downscaling and climate data used
in Rainfall and Rainfall Variability and Impacts of Climate Change on Length of
Growing Period;
Michael Bell at the International Research Institute for Climate and Society (IRI),
who helped extract the Weighted Anomaly of Standardized Precipitation data used
in Rainfall and Rainfall Variability;
ix
Zhe Guo, Melanie Bacou, and Joseph Green at HarvestChoice/IFPRI, who assisted
with data preparation and mapping related to Early Childhood Nutrition and
Health, and Poverty;
Dany Plouffe at McGill University, who assisted with data preparation and analysis
related to Cropland and Pastureland;
Mohamed Fawaz Tulaymat at ICARDA, who helped prepare the Dryland
Systems map;
Verena Henrich at the Institute of Crop Science and Resource Conservation,
University of Bonn, for her research related to Irrigated Areas;
FAO, World Bank, International Institute for Applied Systems Analysis (IIASA),
HarvestChoice, and a large number of agriculturalists, for data and input related to
the Farming Systems analysis;
Michael Morris and Raffaello Cervigni of the World Banks Agriculture and Rural
Development, Africa Region group, who collaborated on the Ruminant Livestock
analysis, which was partially funded by the World Bank as part of a study on
vulnerability and resilience in African drylands; and
Jeffrey Lecksell and Bruno Bonansea of the World Banks Map Design Unit, who
provided the country, lake, and continent boundaries used in the Ruminant
Livestock and World Bank income group maps.
The development of this atlas has been enriched by the support and advice of Deanna
Olney and the main reviewer, Gershon Feder. It also benefited from the copyediting of IFPRIs
Patricia Fowlkes and John Whitehead and proofreading by Heidi Fritschel and Andrew
Marble. The atlas would not be the product it is without the valuable input of IFPRI editor,
Sandra Yin; designer, David Popham; and head of publications, Andrea Pedolsky.
Abbreviations and Acronyms

ACIAR
Australian Centre for International Agricultural Research
AEZ
agroecological zone
AgGDP
agricultural gross domestic product
ASTI
Agricultural Science and Technology Indicators
CAADP
Comprehensive Africa Agriculture Development Program
CCAFS
Climate Change, Agriculture and Food Security
CERES
Crop Environment Resource Synthesis
CGIAR-CSI
CGIAR Consortium for Spatial Information
CIESIN
Center for International Earth Science Information Network
CIMMYT
International Maize and Wheat Improvement Center
CMCC
Euro-Mediterranean Center on Climate Change
CRU-TS
data from the University of East Anglias Climate Research Unit Time Series
CV
coefficient of variation
DHS
Demographic and Health Surveys
DSSAT
Decision Support System for Agrotechnology Transfer
ERGO
Environmental Research Group Oxford
FAO
Food and Agriculture Organization of the United Nations
FAOSTAT
Statistics Division of the FAO and FAOs primary portal for its statistical database
FCC
Soil Functional Capacity Classification System
FTE
full-time equivalent (refers to researchers)
GADM
Global Administrative Boundaries
GAUL
Global Administrative Unit Layers
GCWM
Global Crop Water Model
GDD
growing degree days
GHI
Global Hunger Index
GIS
Geographic Information Systems
GLC2000
Global Land Cover for the year 2000
GMTS
data from the University of Delawares Gridded Monthly Time Series
GNI
average national income
GRUMP
Global Rural-Urban Mapping Project
GYGA-ED
Global Yield Gap Atlas Extrapolation Domain
ha hectares
ICARDA
International Center for Agricultural Research in Dry Areas
ICRAF
World Agroforestry Centre
ICRISAT
International Crops Research Institute for the Semi-Arid Tropics
IDO
intermediate development outcome
IFPRI
IIASA
International Institute for Applied Systems Analysis
ILRI
International Livestock Research Institute
ISRIC
World Soil Information
kg kilograms
km kilometers
LGP
length of growing period
MIRCA
monthly irrigated and rainfed crop areas
mm millimeters
MODIS
Moderate Resolution Imaging Spectroradiometer
xi
MSc
NEPAD
PET
pH
PhD
PPP
R&D
RS
RUE
SPAM
SSA
UMN
UN
UNSALB
US$
VoP
WASP
WDI
WHO
WorldClim
xii
master of science degree

New Partnership for Africas Development
potential evapotranspiration
measure of acidity
doctoral degree
purchasing power parity
research and development
remote sensing
rain-use efficiency
Spatial Production Allocation Model
Africa south of the Sahara
University of Minnesota
United Nations
United Nations Second Administrative Level Boundaries
United States dollars
value of production
Weighted Anomaly of Standardized Precipitation
World Development Indicators
World Health Organization
global climate data layers
Introduction
The Atlas of African Agriculture Research & Development is a multifaceted resource that highlights the ubiquitous nature of smallholder agriculture in Africa; the many factors shaping
the location, nature, and performance of agricultural enterprises; and the strong interdependencies among farming, natural resource stocks and flows, rural infrastructure, and
the well-being of the poor.
Organized around 7 themes, the atlas covers more than 30 topics. Maps illustrate each topic,
complemented by supporting text that discusses the content and relevance of the maps, the
underlying source data, and where to learn more. The atlas is part of an eAtlas initiative that
includes plans for an online, open-access resource of spatial data and tools generated and
maintained by a community of research scientists, development analysts, and practitioners
working in and for Africa.
The atlas got its start in 2009, when Joachim von Braun, a former director general of IFPRI,
was invited to head up the development of the first CGIAR Strategy and Results Framework
(SRF). He asked Stanley Wood, then coordinator of the CGIAR Consortium on Spatial
Information (CSI), to assemble relevant spatial data and analysis to support the analytical
work of the SRF team. Wood first turned to the geographic information system (GIS) specialists at the CGIAR centers to contribute to that effort. Over time researchers at other organizations were invited to contribute.
The many partners and contributors to the atlas share a belief that a better understanding
of the spatial patterns and processes of agriculture research and development in Africa can
contribute to better-targeted policy and investment decisions and, ultimately, to better livelihoods for the rural poor.
To learn more about the eAtlas initiative, visit http://agatlas.org.
xiii
ATLAS
OF AFRICAN AGRICULTURE RESEARCH & DEVELOPMENT
POLITICAL, DEMOGRAPHIC, AND

INSTITUTIONAL CLASSIFICATIONS
Administrative Boundaries 2
Statistical Groupings 4
Public Agriculture R&D Investments 6
Africas Agricultural Research Pool 8
CGIAR Research Program on Dryland Systems 10
Works Cited 12
POLITICAL, DEMOGRAPHIC, AND INSTITUTIONAL CLASSIFICATIONS

Administrative Boundaries
Kate Sebastian
The most common ways to present data for research, demographic, political, and other reporting purposes is by administrative unit or the unit of measure that recognizes the
political boundaries and area of a country. The map shows
Africa divided into nation equivalent (zero-level) units. The
majority of these zero-level units represent countries that
are further divided into smaller subnational (first-level) units,
such as departments or states, which vary in size and number per country.
Drawing boundary lines is often easier said than done.
Discrepancies occasionally occur due to faulty input data or,
on occasion, disputed land areas. An example of this is the
Halaib Triangle, a small area of land over which both Egypt
and Sudan claim sovereignty. Most of the reporting in this
atlas is done at a regional level and both Egypt and Sudan
fall in the northern region so the regional reporting is not
affected. Additionally, South Sudan gained its independence
as a country in 2011 so it is shown separately on the maps
unless the data reported are country-level statistical data that
predate 2011. In such cases South Sudan is not separately
designated (for example, p. 75).
WHY IS THIS IMPORTANT?

As more aid is dispensed and research decisions are made
based on the visualization and mapping of data, it is increasingly important that the boundaries be both accurate
and precise. In creating this atlas, for consistencys sake, it
was imperative that each map use the same administrative boundaries. There are a number of publicly available
worldwide boundary datasets but the Food and Agriculture
Organization of the United Nations (FAO) Global
Administrative Unit Layers (GAUL) is the standard used in
the atlas, because it constantly revises and updates administrative boundaries to present the most up-to-date data available, and it has the highest boundary accuracy rate for the
developing countries of Africa (Figure 1) when compared
to the Global Administrative Boundaries (GADM) and the
UNs Second Administrative Level Boundaries. Using consistent boundaries allows users to easily compare data by
region and even identify patterns. For example, a quick look
at cropland area by region (p. 16) and the average value of
staple food crop production by region (p. 30) shows that
southern Africa not only has the smallest share of total
area devoted to cropland but also the lowest productivity.
FIGURE 1 Accuracy of different administrative boundary

datasets
Number of countries with

correct subnational boundaries
WHAT IS THIS MAP TELLING US?
35
30
25
20
15
10
5
0
GAUL
ADM1
GADM
ADM2
GAUL
ADM1
& GAUL
ADM2
GADM
ADM1
& GADM
ADM2
UNSALB
Source: Adapted from Brigham, Gilbert, and Xu 2013.

Note: GADM=Global Administrative Boundaries; GAUL=Global Administrative Unit Layers (FAO); UNSALB=UN Second Administrative Level Boundaries;
ADM1=First-level administrative boundaries; ADM2=Second-level administrative
boundaries.
Knowing that the same boundaries were used across the

maps gives the reader confidence that these values are based
on the same area totals and thus can be analyzed together.
WHAT ABOUT THE UNDERLYING DATA?

GAUL country boundaries and disputed areas are from
the UN Cartographic Section (FAO 2013). The secondary
boundaries are based on information gathered from both
international and national sources. Data are continuously
being updated and corrected and are released yearly. The
data are licensed strictly for noncommercial use by FAO,
which cannot be held accountable for the accuracy, reliability, or content of the information provided. This is important
to note due to the political nature of the data. Thus, by presenting these boundaries, FAO, and subsequently the organizations involved in this atlas, are not expressing an opinion
concerning the legal status of any area or its authorities or
concerning the delimitation of its boundaries.
WHERE CAN I LEARN MORE?

GAUL 2013 boundaries:
www.fao.org/geonetwork/
GADM boundaries: www.gadm.org
UNSALB boundaries: http://bit.ly/RJ12kD
ATLAS
MAP 1
Country and first-level administrative boundaries
Tunisia
Morocco
Libya
Egypt
Algeria
Mali
Eritrea
Niger
Chad
Senegal
The
Gambia
Guinea
GuineaBissau
Sierra
Leone
Djibouti
Burkina
Faso
Nigeria
Cte
dIvoire Ghana
Liberia
Sudan
Cameroon
Togo Benin
So Tom and Principe
Equatorial
Guinea
Country boundaries
Central African
Republic
Democratic
Republic
of the Congo
Gabon
Ethiopia
South
Sudan
Ug
da
an
Kenya
Rwanda
Burundi
Republic
of Congo
Tanzania
First-level boundaries
Contested areas
Somalia
Angola
Zambia
Namibia
Comoros
ala
wi
a
oz
ue
bi q
M ad
Zimbabwe
Seychelles
c ar
Mauritania
ag a s
Western
Sahara
Botswana
Swaziland
South Africa
Lesotho
Data source: FAO 2013.

Statistical Groupings
Stanley Wood and Kate Sebastian

FAO regional aggregates better reflect similarities in agroecology, language and culture, and market integration opportunities across contiguous constituent countries. World Bank
aggregates reflect similarity in the narrowly defined status
of economic development across geographically dispersed
countries (although the sources of economic growthsuch
as minerals or agriculture or the exploitation of other natural
resources such as timbercan vary widely among countries
in the same economic development category). As shown in
the graphical comparison of different aggregates in Figure 1,
including, for further contrast, total Africa and a split between
landlocked and nonlandlocked country groupings (Map 3),
different regional aggregation schema provide significantly
different insights into the variation of key agricultural performance indicators. Not shown in the maps for reasons of
scale are small African island nations, such as Cape Verde in
western Africa and Reunion in southern Africa. While geographically dispersed, they often face common development
challenges and opportunities (for instance, limited food production potential, sea level rise, and large tourist populations).
The different logical groupings of nations often translate into
formal country associations that represent and promote their
specific common interests. The Convention on Transit Trade
of Land-locked States and the Small Island Developing States
is one example.
Cereal yield (tons per hectare)
6
5
4
TOTAL
INCOMEBASED
REGIONAL
LAND
LOCKED
Cereal yield
Fertilizer consumption
12
10
8
Fertilizer consumption
(10 kilograms per hectare)
The agriculture research and development community

makes extensive use of two primary sources of national
statistics: those compiled by the Food and Agriculture
Organization of the United Nations (FAO), and those
compiled by the World Bank. When presenting summary
statistics across countries in Africa, however, the two organizations use different regional aggregation approaches. FAO
data, accessible through its FAOSTAT portal, are summarized by five geographically contiguous, subregional country groupings: northern Africa, western Africa, middle Africa,
eastern Africa, and southern Africa (Map 1). The World
Bank on the other hand uses an income-based grouping
schema for data accessible through its World Development
Indicators (WDI) portal. Map 2 reflects the World Banks
four categories of average national income per person
(GNI per capita in US dollars): low (<$1,025), lower middle
($1,026$4,035), upper middle ($4,036$12,475), and high
($12,475<) income (World Bank 2013a).
FIGURE 1 Comparing different aggregates,

cereal yields and fertilizer use, 2010
Up
Af
p H
Lo er-m igh rica
i
we id nc
r-m dle om
id inc e*
dl o
e m
Lo inc e
w om
W inc e
e
o
So ster me
ut n A
h f
Ea ern rica
ste Af
M rn A rica
i
No ddle fric
th A a
er fric
n a
A
La fric
n
W d a
ith loc
co ked
as
tli
ne
WHAT ARE THESE MAPS TELLING US?
Data source: FAO 2012a; FAO 2012b; World Bank 2013b.

Note: Because the figure is based on values from 2010, statistics do not include
South Sudan (independent since 2011), in the landlocked countries total. Cereal
crops include barley, buckwheat, canary seed, fonio, maize, millet, oats, quinoa,
rice, rye, sorghum, triticale, and wheat. Fertilizer consumption is based on fertilizer
application for all crops.
*Data unavailable.

FAO compiles and disseminates agricultural production,
consumption, price, input, land use, and related food and
nutrition indicators from country-reported data, while the
World Bank primarily compiles and harmonizes a broader
range of cross-sectoral and macroeconomic data from FAO,
the International Monetary Fund, the International Labour
Organization, the World Health Organization, and other primary sources. Of particular note, however, are the global
responsibilities of FAO and the World Bank to derive, track,
and report on the Millennium Development Goal indicators
of hunger and poverty respectively.

FAOs primary data portal, FAOSTAT, including extensive
metadata descriptions: http://faostat3.fao.org
Other FAO reports and data sets:
http://www.fao.org/publications/
The World Banks WDI data portal: http://bit.ly./1aS5CmL
Extensive WDI poverty-specific datasets:
http://bit.ly/1d7kk9U
ATLAS
MAP 1
FAO regional groups
MAP 2
World Bank income groups

Tunisia
Morocco
Northern
Africa
Algeria
Libya
Egypt
Mauritania
Mali
Western
Africa
Eritrea
Niger
Senegal
Chad
Sudan
Nigeria
Cameroon
Gabon
Democratic
Republic
of the Congo
Rwanda
Republic
of Congo
Low income
Lower-middle income
Upper-middle income
High income
Southern
Africa
Burundi
Ug
an
da
Kenya
Tanzania
Malawi
Angola
Zambia
Zimbabwe
Namibia
oz
am
biq
ue
c ar
Equatorial
Guinea
Ethiopia
ag a s
Benin
Togo
Somalia
South
Sudan
Central African
Republic
M ad
Middle
Africa
The
Gambia Guinea
Cote
Guinea
d'Ivoire
Bissau
Sierra
Leone
Liberia
Ghana
Eastern
Africa
Botswana
Swaziland
South Africa
MAP 3
Djibouti
Burkina
Faso
Lesotho
Landlocked countries
Mali
Niger
Chad
Burkina
Faso
Central African
Republic
South
Sudan
Ethiopia
Uganda
Rwanda
Burundi
Malawi
Landlocked country
Zambia
Zimbabwe
Botswana
Swaziland
Lesotho
R180
G46
B52
R154
G58
B32
R255
G204
B104
R52
G51
B22
R227
G137
B27
R198
G113
B41
Data source: Map 1FAO 2012a; Map 2World Bank 2013b and Lecksell/World Bank 2013; Map 3Authors.

Public Agriculture R&D Investments
Gert-Jan Stads, Nienke Beintema, and Kathleen Flaherty

Growth in public agriculture research and development
(R&D) spending levels in Africa south of the Sahara (SSA)
varied widely from 2008 to 2011 (Map 1). Continent-
wide growth was driven by a handful of larger countries.
However, 13 of the 39 countries for which Agricultural
Science and Technology Indicators (ASTI) data are available experienced negative annual growth in public agricultural R&D spending during 2008/092011.1 Another way of
comparing commitment to public agricultural R&D investment across countries is to measure intensity (Map 2)that
is, total public agricultural R&D spending as a percentage
of agricultural output (AgGDP). Overall investment levels in most countries are still well below the levels required
to sustain agricultural R&D needs. In 2011, SSA as a whole
invested 0.51 percent of AgGDP on average. Just 10 of the
39 countries met the investment target of one percent of
AgGDP set by the African Unions New Partnership for
Africas Development (NEPAD). Some of the smallest countries in Africa, such as Lesotho, Swaziland, Burundi, Eritrea,
and Sierra Leone, have such low and declining levels of
investment that the effectiveness of their national agricultural R&D is questionable. In addition, compared with other
developing regions, agricultural R&D is highly dependent on
funding from donor organizations and development banks
such as the World Bank (Figure 1). This type of funding has
been highly volatile over time, leaving research programs vulnerable and making long-term planning difficult.

A closer look at growth in public agricultural R&D investment levels over time reveals important cross-country differences and challenges. While the intensity ratio of investment
(measured as a share of AgGDP) provides a relative measure
of a countrys commitment to agricultural R&D, monitoring
investments is also key to understanding agriculture R&Ds
contribution to agricultural growth. Research managers and
policymakers can use agricultural R&D spending information
to formulate policies and make decisions about strategic planning, priority setting, monitoring, and evaluation. The data
are also needed to assess the progress of the Comprehensive
Africa Agriculture Development Program (CAADP), which is
designed to boost investments in agricultural growth through
research, extension, education, and training.
1 Due to scale, not all ASTI countries are visible on the maps.
FIGURE 1 Donor funding as a share of total agriculture

R&D funding, 2011
Sudan
Cte d'Ivoire
Kenya
Guinea
Guinea-Bissau
Congo, Dem Rep
Mauritius
Uganda
Tanzania
Ethiopia
Togo
Mauritania
Central Afr Rep
Burundi
Benin
The Gambia
Senegal
Malawi
Liberia
Eritrea
Rwanda
Mozambique
Burkina Faso
Mali
Madagascar
0
10
20
30
40
50
60
70
80
Percent
Source: ASTI 2013.
Note: Donor funding includes loans from development banks and funding from
subregional organizations. Figure excludes countries with donor shares of less than
5 percent.

The data are from primary surveys of 39 countries in SSA
conducted during 20122013 by ASTI and national partners.
ASTI provides comprehensive datasets on agricultural R&D
investment and capacity trends and institutional changes
in low- and middle-income countries. The datasets are
updated at regular intervals and accessible online.

ASTI datasets, publications, and other outputs by country:
www.asti.cgiar.org/countries
ASTI methodology and data collection procedures:
www.asti.cgiar.org/methodology
ATLAS
Ethiopia
Central African
Republic
Ug
Gabon
Rwanda
Burundi
Republic
of Congo
Kenya
Republic
of Congo
Tanzania
ue
bi q
Zambia
Namibia
za
Zimbabwe Mo
Botswana
3
Swaziland
No data or non-ASTI country
South Africa
Lesotho
Data source: ASTI 2013.

Notes: AgGDP=agricultural output. Intensity of agricultural R&D spending=public
agricultural R&D spending per $100 of agricultural output.
Ethiopia
Democratic
da
an Kenya
Republic
Ug
of the Congo
Rwanda
Burundi
Tanzania
Malawi
$ of R&D spending
per $100 AgGDP
0
0.5
1.00
2.00
Sudan
Central African
Republic
Gabon
Malawi
Annual spending
growth rate (%)
23
3
da
an
Chad
Burkina
The
Faso
Gambia
Guinea
Nigeria
GuineaCte
Bissau
d'Ivoire
Sierra
Leone
Liberia Ghana Togo Benin
Nigeria
Cte
d'Ivoire
Sierra
Leone
Eritrea
Mali
ue
bi q
Zambia
Namibia
za
Zimbabwe Mo
c ar
Burkina
Faso
c ar
Guinea
Senegal
Sudan
ag a s
Chad
ag a s
The
Gambia
Mauritania
Eritrea
Mali
M ad
Mauritania
Senegal
MAP 2Intensity of agriculture R&D spending, 2011
M ad
MAP 1 Change in public agriculture R&D spending levels,

20082011
Botswana
Swaziland
South Africa
Lesotho
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE

Africas Agricultural Research Pool
Nienke Beintema, Gert-Jan Stads, and Kathleen Flaherty

Absolute levels of staffing in public agriculture research
and development (R&D) vary considerably across the
39 countries in Africa south of the Sahara participating in the Agricultural Science and Technology Indicator
(ASTI) survey (Map 1). In 2011, Ethiopia, Ghana, Kenya,
Nigeria, South Africa, Sudan, and Tanzania each employed
more than 500 full-time equivalent (FTE) researchers.
In contrast, 11 countries employed fewer than 100 FTE
researchers each.1 Despite recent challenges, many western African countries have maintained relatively large
pools of well-qualified researchers (those holding PhD
and MSc degrees) (Map 2). In contrast, less than half of
researchers in Botswana, the Democratic Republic of the
Congo, Eritrea, Ethiopia, Guinea, Guinea-Bissau, Lesotho,
Liberia, Mozambique, and Zimbabwe hold graduate
degrees. Map 3 shows the number of FTE researchers per
100,000 people who are economically active in agriculture. While the overall average for ASTI countries is 7 FTE
researchers per 100,000, only Botswana, Cape Verde, Gabon,
Mauritius, Namibia, Nigeria, and South Africa each employ
more than 20 FTEs per 100,000 agriculture sector workers.

There is growing concern about the ability of African agriculture research and development (R&D) systems to respond
to current and emerging development challenges. Some
of Africas smallest countries have such low, and in a few
instances, declining levels of researcher numbers that the
effectiveness of their national agricultural R&D systems is
questionable. Structural problems also persist in the age
and sex composition of R&D personnel (Figure 1 provides a
national example), where the limitations of an aging research
workforce and knowledge base are exacerbated by the low
participation of female researchers (especially when compared to their much broader participation in the sector as
farmers, farm workers, and traders). Furthermore, despite
stable growth in the number of agricultural researchers,
many research agencies experienced high staff turnover as a
1 Due to scale, not all ASTI countries are visible on the maps.
FIGURE 1 Age and sex structure of agricultural R&D

staff: Senegalese Agricultural Research Institute, 2008
5559
Female
Male
5054
Age
4549
4044
3539
3034
28
21
14
7
0
7
14
Number of researchers
21
28
Source: Sne et al. 2012.
consequence, in part, of researchers retiring from the workforce (Beintema and Stads 2011). Aging scientist populations
and the deterioration of average degree levels in many countries imply a chronic erosion of domestic innovation capacity. Ongoing monitoring of national agriculture research
capacity can contribute to the formulation of appropriate responses.

Underlying primary data are from 39 national surveys conducted during 20122013 by the ASTI initiative and national
partners. ASTI generates and curates comprehensive and
comparable agriculture R&D institutional, investment, and
capacity data for low- and middle-income countries. The
datasets are periodically updated and are accessible online.

ASTI datasets, publications, and other outputs by country:
www.asti.cgiar.org/countries
ASTI methodology and data collection procedures:
www.asti.cgiar.org/methodology
Other ASTI resources: www.asti.cgiar.org/about
ATLAS
Democratic
Republic
of the Congo
Ug
Rwanda
Burundi
Republic
of Congo
da
an
ue
bi q
Zambia
m
ag a s
Botswana
Swaziland
South Africa
Rwanda
Burundi
Ug
da
an
Kenya
Tanzania
Malawi
Percent of total
M ad
za
Zimbabwe Mo
Ethiopia
Democratic
Republic
of the Congo
Republic
of Congo
Tanzania
Sudan
Central African
Republic
Gabon
Malawi
Namibia
Chad
Kenya
Total FTE researchers
Eritrea
Mali
Burkina
The
Faso
Gambia
Guinea
Nigeria
GuineaBissau
Sierra
Leone
Ethiopia
Central African
Republic
Gabon
Senegal
Sudan
0
50
75
90
Lesotho
Namibia
za
Zimbabwe Mo
ue
bi q
ag a s
Chad
Burkina
The
Faso
Gambia
Guinea
Nigeria
GuineaCte
Bissau
d'Ivoire
Sierra
Leone
0
100
200
500
Mauritania
Eritrea
Niger
M ad
Mali
c ar
Mauritania
Senegal
MAP 2 Share of agricultural researchers with

postgraduate degrees, 2011
c ar
MAP 1Number of agricultural researchers, 2011
Botswana
Swaziland
Lesotho
MAP 3Concentration of agricultural researchers, 2011
Eritrea
Niger
Chad
Burkina
The
Faso
Gambia
Guinea
Nigeria
GuineaCte
Bissau
d'Ivoire
Sierra
Leone
Democratic
Republic
of the Congo
Rwanda
Burundi
Republic
of Congo
Ug
da
an
Kenya
Tanzania
Malawi
ue
bi q
Zambia
Namibia
za
Zimbabwe Mo
c ar
2
10
20
40
Ethiopia
Central African
Republic
Gabon
FTEs per
100,000 farmers
Sudan
ag a s
Mali
M ad
Mauritania
Senegal
Botswana
Swaziland
South Africa
Lesotho
Data source: Maps 1 and 2ASTI 2013; Map 3 ASTI 2013 and FAO 2013.
Notes: Maps 1 and 3FTE=full-time equivalent. FTE values take into account only the proportion of
time spent on research and development; Map 2Researchers with postgraduate degrees earned PhDs
or MScs; Map 3Farmers include all agricultural sector workers.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE

CGIAR Research Program on Dryland Systems
Chandrashekhar Biradar

The map shows the distribution of dryland agricultural
production systems (also known as the CGIAR Research
Program on Dryland Systems) in Africa. Dryland systems are
characterized by low and erratic precipitation, persistent
water scarcity, extreme climatic variability, high susceptibility to land degradation, including desertification, and higher
than average loss rates for natural resources, such as biodiversity. The lack of water is the main factor that limits profitable
agricultural production. Dryland systems consist of combinations of plant and animal species and management practices
farmers use to pursue livelihood goals based on several factors including climate, soils, markets, capital, and tradition.
Dryland Systems is a multidisciplinary research program that
aligns the research ofCGIAR research centers and partners. It
aims to tackle complex development issues in two key strategic research themes known as intermediate development
outcomes (IDOs). The first IDO focuses on low-potential and
marginal drylands and developingstrategies and tools to
minimize risk and reduce vulnerability. The second IDO
focuses on higher-potential dryland regions andsupporting
sustainable intensification of agricultural production systems.
Within each large target area, a number of representative
action sites and complementary satellite sites serve as test
sites where most of the research will be conducted. These
siteswhich include the Kano-Katsina-Maradi Transect
in Nigeria and Niger; Wa-Bobo-Sikasso Transect in Ghana,
Burkina Faso, and Mali; Tolon-K and Cinzana along West
African Sahel and dryland savannas in Ghana and Mali; the
Nile Delta in Egypt; Bni Khedache-Sidi Bouzid inTunisia;
the Ethiopian Highlands; and Chinyanja Triangle in Malawi,
Zambia, and Mozambiquewere identified based on criteria relating to aridity index, length of growing period, market
access, hunger and malnutrition, poverty, environmental risk,
land degradation, and demography.

The goal of the Dryland Systems research programis to identify and develop resilient, diversified, and more productive
10
TABLE 1
Dryland Systems sites in Africa, 2013
Action Sites
IDO1
IDO2
32,861,151
60,865,568
Population
924,092
18,621,053
Households
184,818
3,724,211
Area (ha)
Source: Author.
Note: IDO=intermediate development outcomes.
combinations of crop, livestock, rangeland, aquatic, and

agroforestry systems that increase productivity, reduce hunger and malnutrition, and improve quality of life for the rural
poor. The research program aims to reduce the vulnerability of rural communities and entire regions across the worlds
dry areas by sustainably improving agricultural productivity.
The map provides a starting point for implementing interventions for intermediate development outcomes. It also can
help researchers extrapolate from the research outcomes at
action sites to target areas and scale up better interventions
to target regions over time.

The Remote Sensing (RS)/Geographic Information Systems
(GIS) Units of the participating CGIAR centers characterized
dryland systems to delineate target areas, action sites, and
complementary satellite sites, using various spatial layers,
such as aridity index (p. 55), length of growing period
(p. 57), access to markets (p. 66), environmental risk, land
degradation, and additional criteria from regional and
representative target region perspectives (CGIAR 2012).

Dryland Systems: http://drylandsystems.cgiar.org
ICARDA Geoinformatics: http://gu.icarda.org
Dryland Systems and Other CGIAR Research Programs:
http://bit.ly/1eQnJdC
ATLAS
MAP 1
Dryland Systems action sites and target research areas
Target areas
IDO1
IDO2
Sites
IDO1 action site
IDO1 satellite site
IDO2 action site
IDO2 satellite site
Data source: GeoInformatics Unit/ICARDA 2013.

Note: IDO=intermediate development outcomes. Action sites are representative areas of major
widespread agroecosystems where initial intervention takes place to identify best approaches
and top priorities for scaling out to large areas (target regions). Satellite sites are complementary
(back-up) action sites.
11

Works Cited
ADMINISTRATIVE BOUNDARIES
AFRICAS AGRICULTURAL RESEARCH POOL
Brigham, C., S. Gilbert, and Q. Xu. 2013. Open Geospatial Data: An Assessment of Global
Boundary Datasets. The World Bank. Accessed November 11, 2013.
http://bit.ly/1cxvTnE.
FAO (Food and Agriculture Organization of the United Nations). 2013. Political Boundaries:
Global Administrative Unit Layers (GAUL). www.fao.org/geonetwork/.
ASTI (Agricultural Science and Technology Indicators). 2013. ASTI database. Accessed
January 17, 2013. www.asti.cgiar.org/data/.
Beintema, N., and G.-J. Stads. 2011. African Agricultural R&D in the New Millennium: Progress
for Some, Challenges for Many. Food Policy Report 24. Washington, DC: International
Food Policy Research Institute.
FAO. 2013. FAOSTAT Database. http://faostat.fao.org/site/291/default.aspx.
Sne, L., F. Liebenberg, M. Mwala, F. Murithi, S. Sawadogo, and N. Beintema. 2011. Staff
Aging and Turnover in African Agricultural R&D: Lessons from Five National Agricultural
Research Institutes. Conference Working Paper No. 17. Washington, DC: International
Food Policy Research Institute and Forum for Agricultural Research in Africa.
STATISTICAL GROUPINGS
FAO (Food and Agriculture Organization of the United Nations). 2012a. FAOSTAT database. Accessed October 15, 2012. http://faostat.fao.org/site/291/default.aspx.
. 2012b. FAOSTAT database. Accessed December 2, 2012.
http://faostat.fao.org/site/291/default.aspx.
Lecksell, J./ World Bank. 2013. Personal communication regarding world boundaries for
2013, Nov. 26. Lecksell is lead World Bank cartographer.
World Bank. 2013a.World Development Indicators 2013. Washington, DC: World Bank.
http://bit.ly/1pW4Rzc.
. 2013b. World Development Indicators (Income Levels). Accessed December 22,
2013. http://bit.ly/1aS5CmL.
PUBLIC AGRICULTURE R&D INVESTMENTS

ASTI (Agricultural Science & Technology Indicators). 2013. ASTI Countries. Accessed
October 14, 2013. www.asti.cgiar.org/countries.
12
CGIAR RESEARCH PROGRAM ON DRYLAND SYSTEMS

CGIAR. 2012. Research Program on Dryland Systems. Accessed November 19, 2013.
http://drylandsystems.cgiar.org.
ICARDA (International Center for Agricultural Research in the Dry Areas). 2013.
GeoInformatics Unit. http://gu.icarda.org.
ATLAS
Footprint of Agriculture
Farming Systems of Africa 14
Cropland and Pastureland 16
Irrigated Areas 18
Cereal Crops 20
Root Crops 22
Livestock and Mixed Crop-Livestock Systems 24
Ruminant Livestock 26
Cropping Intensity 28
Land Productivity for Staple Food Crops 30
Works Cited 32
13
FOOTPRINT OF AGRICULTURE
Farming Systems of Africa
Christopher Auricht, John Dixon, Jean-Marc Boffa, and Dennis Garrity

Broadly similar farming systems share recognizable livelihood
patterns and similar development pathways, infrastructure, and policy needs. Delineating major farming systems
provides a framework to guide the development and targeting of strategic agricultural policies and interventions to
reduce poverty and promote the adoption of more sustainable land use practices. This classification can help policymakers and scientists target institutional innovations and
technologies to specific farming systems, thereby focusing
planning, policies, and research. In this respect, high potential farming systems with good market access might benefit
most from improved maize, cowpeas, and dairy, while drier
areas might benefit from improved sorghum, millet, and livestock, because these contrasting farming systems offer different ways to improve livelihoods. Similarly, fertilizer policies
should take into account the different nutrient requirements
and markets of various crops in different farming systems.

Farming systems are defined based on: available natural resources (including water, land area, soils, elevation,
and length of growing period); population; cropping and
14
60
People (millions)
Populations within the same farming system share similar

farming practices and livelihood strategies. As the map shows,
many farming systems in Africa exhibit a strong geographical
pattern, extending across northern Africa and Africa south
of the Sahara (SSA), reflecting a mix of factors, including
climate, soils, and markets. In SSA, 16 percent of land area
is dominated by the maize mixed farming system, mostly
in the eastern, central, and southern regions. This farming
system is home to nearly 100 million rural people, of whom
58 million live on less than $1.25 a day (Figure 1), representing 23 percent of the total rural poor in SSA. The highland
areas of eastern and southern Africa feature smaller fragmented systems, such as the highland perennial and highland mixed systems that cover just 2 percent of the area, but
are home to 11 and 6 percent, respectively, of SSA rural poor.
A large share of the rural poor live in the agropastoral farming system (18 percent), root and tuber crop system (11 percent), and cereal-root crop mixed system (10 percent), which
combined cover more than one-third of SSAs area.
FIGURE 1 Rural poor living on $1.25/day

by farming system, Africa south of the Sahara, 2010
50
40
30
20
10
0
M
aiz
em
A
Hi gr ixe
o
gh
pa d
l
Ro and sto
o
p ra
Ce t an ere l
nn
re
d
alt
i
ro ube al
ot
rc
r
cr
Hu
o op
m Hig p m
id
lo hlan ixed
wl d
an m
d ixe
tre d
ec
ro
Pa p
sto
Fis ral
h
Fo -ba
re sed
stba
se
Pe Irri d
ga
re
Ar nn
te
d
id
i
pa al m
sto ixe
ra d
l-o
as
es
Data source: Dixon, Boffa, and Garrity 2014; Azzarri et al. 2012; UN 2013.
Note: See glossary for definitions of specific farming systems. Poverty data
calibrated to 2010.
pasture extent; the dominant pattern of farm activities and

household livelihoods; and access to markets. The spatial
characterization of African farming systems used data on
agroecological and socioeconomic variables. The two main
spatial variables were length of growing period (FAO/IIASA
2012) and distance to markets (HarvestChoice 2011; Map 1,
p. 67), supplemented by data on population and poverty,
elevation, soils and irrigation, crop and livestock patterns,
productivity, and change over time (Dixon et al. 2014; FAO
2013ae). A multidisciplinary team of experts for each farming system identified system characteristics, emergent properties, drivers of change and trends, and priorities. This work
updates and expands the analysis of the African farming systems in the World Bank and FAO farming systems and poverty assessment (Dixon et al. 2001).

Farming Systems and Poverty: Improving Farmers Livelihoods
in a Changing World. Dixon et al. 2001.
http://bit.ly/1dDekBW
Understanding African Farming Systems: Science and Policy
Implications. Food Security in Africa: Bridging Research into
Practice. Garrity et al. 2012. http://bit.ly/1h8lmGJ
ATLAS
MAP 1
Farming systems of Africa
Maize mixed
Agropastoral
Highland perennial
Root and tuber crop
Cereal-root crop mixed
Highland mixed
Humid lowland tree crop
Pastoral
Fish-based
Forest-based
Irrigated
Perennial mixed
Arid pastoral-oasis
North Africa dryland mixed
North Africa rainfed mixed
North Africa highland mixed
Data source: Dixon, Boffa, and Garrity 2014.

Note: See glossary for definitions of specific farming systems.
15
Cropland and Pastureland
Navin Ramankutty

These maps of cropland and pastureland provide critical
pieces of information used to analyze food security and
agricultures environmental impact. More accurate assessments of the land under cultivation and areas potentially
available for expansion could help improve food security.
For instance, in Africa south of the Saharaone of the only
regions in the world where increases in food production
have not kept pace with population growththe land area
suitable for cultivation is estimated to be nearly five times
what is currently in production. Knowledge of pasturelands
is similarly vital to food security because livestock provide
not only a source of food but also income, insurance, soil
nutrients, employment, traction (for instance, plowing),
and clothing (Thornton and Herrero 2010). However, both
grazing and planting also contribute to environmental degradation (Foley et al. 2005) and already have modified a
large part of the African continent. Overgrazing contributes to land degradation, further diminishing soil health,
plant productivity and diversity, and by extension, livestock
production. Grazing is also a significant source of methane
emissions, a potent greenhouse gas that contributes to climate change.
16
Cropland and pastureland by region, c. 2000

Share of total
(%)
Total area
(000 sq km)
Share of total
(%)
Region
Pasture area
(000 sq km)
Crop Area
Share of total
(%)
Map 1 shows the extent of cropland, and Map 2 shows the

extent of pastureland circa 2000. The values are presented
as a percentage of each ~100 km2 grid cell. Pastureland covers one-quarter of the African continent (Table 1) and dominates the landscape in the Sahel and Sudano-Sahelian regions
in the west, the Maghreb, much of eastern and southern
Africa, and western Madagascar. The only portions of the
continent not grazed are those that are too hot and too dry,
such as the Sahara, and the tropical rain forests of the Congo
Basin. Cropland covers approximately 7 percent of the continent. Western Africa has the greatest proportion at 39 percent. High concentrations of cropland (60 percent) can
be found along the Mediterranean coast in the Nile Valley,
Nigeria, the Ethiopian highlands, the Rift Valley north and
west of Lake Victoria, and South Africa near Cape Town
and north of Lesotho. Low-to-moderate cropland intensity
(2060 percent) extends from Nigeria to Senegal and can be
found in parts of Sudan, and scattered throughout southeastern Africa.
TABLE 1
(000 sq km)
Eastern Africa
501
23.4
2,404
31.3
6,172
20.9
Middle Africa
249
11.6
1,144
14.9
6,448
21.8
Northern Africa
374
17.5
1,603
20.9
8,266
27.9
Southern Africa
172
8.0
1,380
18.0
2,683
9.1
Western Africa
842
39.4
1,149
15.0
6,031
20.4
2,138
100.0
7,680
100.0
29,599
100.0
Total
Data source: Ramankutty et al. 2008 and FAO 2012.

Note: sq km=square kilometers.

The distribution and intensity of croplands and pastures
are expressed as a percentage of the area within each
~100 km2 grid cell. The maps represent arable land and permanent crops and permanent meadows and pastures,
respectively, as defined by the Food and Agriculture Organi
zation of the United Nations (FAO 2013). Data for both maps
derive from integrating administrative-level agricultural statistics with global land cover classification data from satellite remote sensing using a statistical data fusion method
(Ramankutty et al. 2008). The agricultural statistics for Africa
came mainly from FAOs national statistics (FAOSTAT 2012),
supplemented with subnational statistics for Nigeria and
South Africa. Two different sources of satellite-based land
cover classification data were merged: the MODIS land cover
dataset from Boston University (Friedl et al. 2010) and the
GLC2000 dataset (Bartholom and Belward 2005) from the
European Commission (both at 1 km spatial resolution).

Download crop and pasture data at EarthStat:
www.earthstat.org
Farming the Planet. Part 1: The Geographic Distribution of
Global Agricultural Lands in the Year 2000.
Ramankutty et al. 2008: http://bit.ly/1ctE7Nf
ATLAS
MAP 1
Cropland, c. 2000
MAP 2
Pastureland, c. 2000
Percent
Percent
0
0<
20
40
60
80
100
0
0<
20
40
60
80
100
Data source (all maps): Ramankutty et al. 2008.

Notes: All values are expressed as a percentage of the area within each ~100 km2 grid cell. Cropland=arable land and
permanent crops; pastureland=permanent meadows and pastures, as defined by the Food and Agriculture Organization
of the United Nations (FAO 2013).
17
Irrigated Areas
Stefan Siebert and Karen Frenken

Total area equipped for irrigation in Africa is 13.5 million
hectares (ha) of which 11.5 million ha are actually under
irrigation (Figure 1). The map shows the countries with
the largest amount of area equipped for irrigation are
Egypt (3.5 million ha), Sudan and South Sudan (1.9 million
ha), South Africa (1.5 million ha), and Morocco (1.5 million ha). All of these countries face arid climate conditions.
In Madagascar where it is more humid, rice is cultivated
on about 1 million ha of irrigated land. These six countries account for almost 60 percent of the area equipped
for irrigation in Africa. The regions with the highest density of irrigated land (50 percent or greater of the grid
cell)1 are located mainly in northern Africa in the Nile
River Basin (Egypt, Sudan) and in the countries next to
the Mediterranean Sea (Morocco, Algeria, Tunisia, Libya).

Since the beginning of crop cultivation, irrigation has been
used to compensate for the lack of precipitation. In rice
cultivation, irrigation also controls the water level in the
fields and suppresses weed growth. Crop yields are higher
and the risk of crop failures is lower in irrigated agriculture.
Because the risk of drought stress is lower on irrigated land,
farmers are more likely to spend on other inputs like fertilizers. Irrigation may also increase cropping intensity (p. 28),
allowing farmers to cultivate several crops per year on the
same field. It is important, therefore, when assessing crop
productivity and food security, to consider the availability of
irrigation infrastructure.
Irrigation represents the largest use of freshwater in
Africa. Many dams were constructed to improve the supply of irrigation water, thereby modifying river discharge and
increasing evaporation from artificial lakes. Extraction of
groundwater for irrigation is increasingly of concern, because
it has lowered groundwater tables in important aquifers. Use
of irrigation results in an increase of evapotranspiration and
reduces the lands surface temperature. Information on the
extent of irrigated land is therefore also important for hydrological studies and regional climate models.

The map shows the area equipped for irrigation as a percentage of a 5 arc-minute grid cell. It was derived from
1 Each cell measures approximately 100 km2 or 10,000 hectares at the equator.
18
FIGURE 1 Area equipped for irrigation and area actually

irrigated per region, c. 2005
9
8
Irrigated area (million ha)
Area equipped
for irrigation
Area actually
irrigated
7
6
5
4
3
2
1
0
Eastern
Africa
Middle
Africa
Northern
Africa
Southern
Africa
Western
Africa
Data source: Siebert et al. 2013a and FAO 2012.
version 5 of the Digital Global Map of Irrigation Areas

(Siebert et al. 2013a). The map was developed by combining subnational irrigation statistics for 441 administrative
units derived from national census surveys and from reports
available at the Food and Agriculture Organization of the
United Nations and other international organizations with
geospatial information on the position and extent of irrigation schemes. Statistics for the year closest to 2005 were
used if data for more than one year were available. Geospa
tial information on position and extent of irrigated areas
was derived by digitizing a large number of irrigation maps
derived from inventories based on remote sensing (Siebert
et al. 2013b).

Global Map of Irrigation Areas (Version 5):
http://bit.ly/1eHpDex
Update of the Digital Global Map of Irrigation Areas to
Version 5. Siebert et al. 2013b: http://bit.ly/1cM6bip
Development and Validation of the Global Map of
Irrigation Areas. Siebert et al. 2005.
ATLAS
MAP 1
Extent of irrigated areas, c. 2005
Morocco
Tunisia
Libya
Algeria
Egypt
Libya
Chad
Sudan
Ethiopia
Percent irrigated
0
0<
1
5
10
20
35
50
75
100
Data source: Siebert et al. 2013a.

Note: The percent values represent the share of each 100 km2 cell that is
equipped for irrigation.
19
Cereal Crops
Ulrike Wood-Sichra

Cereals account for 50 percent of the average daily caloric
intake in Africa. Wheat and rice are particularly important,
accounting for 30 percent and 16 percent of cereal calories
consumed, respectively. Cereal production in Africa is substantial, but it is not enough to meet demand; the continent
must import about 55 percent of consumed wheat and more
than 30 percent of consumed rice (FAO 2012). Understanding
where half of the continents calories (both vegetal and animal) are grown, and how intensively, is vital to increasing
productivity and enhancing food security. Identifying areas
where new or improved rice- and wheat-growing technologies could have the most impact can also aid in making the
continent less dependent on imports.
Area harvested of top five cereal crops
35
Maize
Millet
Rice, paddy
30
Million hectares
Cereals are grown in all of Africa except for desert and

forested areas. The cereal area is about 30 percent maize,
23 percent sorghum, 21 percent millet, 9 percent wheat
(Maps 14), and 9 percent rice. Maps 13 show that maize
is prevalent throughout Africa and the densest areas for sorghum and millet, with more than 3,000 hectares per cell,1 are
just south of the Sahel. Wheat (Map 4) is grown in high concentrations in northern Africa, with sparser areas in eastern
and southern Africa. In the last 50 years, the harvested areas
of maize, millet, and sorghum each doubled from a base of
1015 million hectares to 2030 million hectares (Figure 1).
Rice areas have nearly quadrupled, from 2.8 to 9.3 million
hectares. Yields have notably climbed for maize and wheat
during the same period, rising from 0.7 to 2.3 metric tons
per hectare for wheat and doubling from 1.0 to 2.0 for maize
(Figure 2). Rice yields have increased by more than half, from
about 1.5 to 2.5 metric tons per hectare. Millet and sorghum
yields show little change (FAO 2012).
FIGURE 1
25
Sorghum
Wheat
20
15
10
5
0
1961
1968
1975
1982
1989
1996
2003
2010
1996
2003
2010

FIGURE 2
Yield of top five cereal crops
3.0
2.5
Metric tons/hectare
Maize
Millet
Rice, paddy
Sorghum
Wheat
2.0
1.5
1.0
0.5
0.0
1961
1968
1975
1982
1989

Note: One metric ton=1,000 kilograms.
of the United Nations (FAO 2012), and data on production

systems within each country.

The maps are based on area harvested per cell, calculated
using the Spatial Production Allocation Model (SPAM) 2000
(You et al. 2012). The model uses many datasets, including
land cover recorded by satellites, crop suitability maps under
various water regimes and production systems, irrigation
maps (p. 19), subnational crop statistics from each country,
country totals from the Food and Agriculture Organization
20

SPAM: The Spatial Production Allocation Model:
http://mapspam.info
Statistics Division database: http://faostat3.fao.org
ATLAS
Cereal crop area harvested, 2000

MAP 1Maize
MAP 2Sorghum
Hectares
0
0<
10
250
500
1,000
3,000
MAP 3Millet
Hectares
0
0<
10
250
500
1,000
3,000
MAP 4Wheat
Hectares
0
0<
10
250
500
1,000
3,000
Hectares
0
0<
10
250
500
1,000
3,000
Data source (all maps): You et al. 2012.

Note: The values on the maps represent the number of hectares harvested per 100 km2 cell.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
21
Root Crops
Ulrike Wood-Sichra
The area devoted to harvest root crops in Africa has grown
significantly over the last 50 years. Cassava area has more
than doubled, from 5.5 million to 12 million hectares (ha);
sweet potato area has more than quintupled, from
600,000 to 3.3 million ha; and potato area has grown more
than six-fold, from 250,000 to 1.8 million ha. Cassava and
sweet potatoes continue to be among the most important root crops in Africa, with cassava occupying about half
of the root crops area and sweet potatoes about 14 percent. South of the Sahel, cassava and sweet potatoes are
grown in similar areas (Maps 1 and 2). Both are grown intensively, with 1,000 or more ha per cell,1 in the southeast corner of Nigeria, in the eastern part of Uganda, and in Rwanda
and Burundi. Potatoes are becoming a more important
part of Africas crop mix, although they currently account
for only 8 percent of the harvested area and are grown in
just a few African countries (Map 3). While harvested area
of root crops has expanded considerably since 1961, yields
per hectare have increased significantly for only some crops
(Figure 1). Cassava yields have notably improved by about
80 percent, from less than 6 to roughly 10 metric tons per
hectare. Potato yields have also fared well, increasing by
about half from about 8 to 12 metric tons per hectare. Taro
and yam yields grew more modestly, by 41 percent and 26
percent to 6 and 10 metric tons per hectare, respectively.
Sweet potato yields, however, have hovered around 5 metric
tons per hectare for decades and even shown a slight downward trend over the past 30 years (FAO 2012).

Africa needs to improve yields and the share of nutrient-rich
roots and tubers in the diet of its growing population.
Roots and tubers contribute only about 13 percent of the
calories in the average Africans diet, which is a smaller portion than other staples. But roots, especially cassava, are
insurance crops that increase food security because they
can be left in the ground until needed. Nearly all the sweet
potato crop (85 percent) is destined for human consumption. But cassava is also important as fodder, and more than
a third produced goes to animal feed. Most of the roots and
tubers consumed are grown locally. Thus, policymakers and
22
FIGURE 1
Yield of top five root crops
14
12
Metric tons/hectare
Cassava
Taro (cocoyam)
Potatoes
Yams
Sweet potatoes
10
8
6
4
2
1961 1967 1973 1979 1985 1991 1997 2003 2009

Note: One metric ton=1,000 kilograms.
agricultural experts can use the maps to identify areas that

might benefit from larger harvests of roots and tubers, and
by extension, improve nutrition at the local level.

The maps are based on area harvested per cell, calculated
by the Spatial Production Allocation Model (SPAM) 2000
(You et al. 2012). The model uses many datasets, including land cover recorded by satellites, crop suitability maps
under various water regimes and production systems, irrigation maps, subnational crop statistics from each country,
country totals (FAO 2012), and data on production systems
within each country.

SPAM: The Spatial Production Allocation Model:
http://mapspam.info
statistical database: http://faostat3.fao.org
ATLAS
Root crop area harvested, 2000

MAP 1Cassava
MAP 2
Hectares
0
0<
10
250
500
1,000
3,000
Sweet potato
Hectares
0
0<
10
250
500
1,000
3,000
MAP 3Potato
Hectares
0
0<
10
250
500
1,000
3,000
Data source (all maps): You et al. 2012.

Note: The values on the maps represent the number of hectares harvested per 100 km2 cell.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
23
Livestock and Mixed Crop-Livestock Systems
Philip Thornton

Livestock-producing agricultural systems cover 73 percent
of Africa and stretch across several climates (Map 1). To
some extent, these climates determine what type of farming
is practiced. In Africa, livestock-producing systems are broken into two main categories: livestock and mixed crop-livestock. These systems exist in three common African climates:
arid/semiarid, humid/subhumid, and temperate/tropical
highlands. Livestock systems are most prevalent on grazing lands in arid climates that cover large swaths of Africa.
Mixed crop-livestock farming systems are either rainfed or irrigated. Rainfed systems are much more common
(although areas of Sudan and Egypt have important irrigated mixed systems that present different opportunities
and constraints). There are many mixed crop-livestock systems throughout western Africa, eastern Africa, and parts of
southern Africa. The Congo Basin, in central Africa, is mostly
forest, with some savanna and cropland at its outer edges.
As a result, the Basin is home to a small number of livestock
systems relative to the rest of the continent and only a smattering of mixed crop-livestock systems.

Many studies have found the influences of crop and livestock production vary considerably, not only regionally
but also according to production system (Robinson et al.
2011). Globally, but particularly in Africa and Asia, crops
and livestock are often interdependent and influence farmer
households and livelihoods in a number of ways. Detailed
knowledge of crop and livestock systems and their distribution allows researchers to measure impacts on everything
from the environment to livestock disease risk. For example, viewing the livestock density by type and region helps
researchers measure the level of environmental impact
(Table 1). Classification of agricultural systems can also provide a framework for predicting the evolution of the agricultural sector in response to changing demography and
associated shifts in food demand, land use (for example,
competition for land from food, feed, and biofuel production), and climate.

The systems classification is based on Ser and Steinfeld
(1996). In livestock systems, more than 90 percent of dry
24
TABLE 1
Livestock density by region, 2005

TYPE OF LIVESTOCK
REGION
average number/km2
Cattle
Sheep
Goat
Northern Africa
10
Middle Africa
Eastern Africa
14
Western Africa
10
12
Southern Africa
11
AFRICA
Data source: Robinson et al. 2011 and FAO 2012.
matter fed to animals comes from rangelands, pastures,

annual forages, and purchased feeds, and less than 10 percent of the total value of production (VoP) comes from
nonlivestock farming activities. Mixed crop-livestock farming systems are systems in which more than 10 percent
of the dry matter fed to animals comes from crop by-
products (for example, stubble) or more than 10 percent
of the total VoP comes from nonlivestock farming activities. The systems were mapped using various mapped data
sources, including land cover data, irrigated areas, human
population density, and length of growing period (LGP).
The climate categories are defined as follows: arid/semiarid has an LGP 180 days; humid/subhumid has an LGP
> 180 days; and the temperate/tropical highlands climate
is based on specific LGP, elevation, and temperature criteria. The systems classifications have several weaknesses,
including differences in estimates of the amount of Africas
cropland, depending on the data used, thus, there is some
uncertainty in identifying the mixed crop-livestock systems. Researchers are now using other data sources to
break down the mixed systems of the Ser and Steinfeld
classification by dominant food and feed crop categories
(Robinson et al. 2011).

Farming Systems and Poverty: Improving Farmers Livelihoods
in a Changing World. Dixon et al. 2001.
Global Livestock Production Systems. Robinson et al. 2011.
ATLAS
MAP 1
Livestock production systems by climate zone
Livestock-grazing
Rainfed-arid/semiarid
Rainfed-humid/subhumid
Rainfed-temperate/tropical highlands
Mixed crop-livestock
Rainfed-arid/semiarid
Rainfed-humid/subhumid
Rainfed-temperate/tropical highlands
Irrigated
Other
Urban areas
Nonlivestock vegetated areas
Data source: Robinson et al. 2011.

Note: The mixed categories represent a mix of crop and livestock systems.
25
Ruminant Livestock
Timothy Robinson, William Wint, Giulia Conchedda, Guiseppina Cinardi, and Marius Gilbert
Ruminant livestock are raised across large parts of Africa
where environmental conditions allow (Maps 14). Cattle,
sheep, and goats are the most widespread, while camels are
restricted to drier areas, particularly in the Horn of Africa and
the arid parts of western Africa. These maps of ruminant distribution should, however, be used in conjunction with the
livestock production systems map (p. 25) to better understand the systems and climate zones where ruminant livestock are found. The role of livestock varies greatly depending
on the production system. The heavily forested areas and
hyperarid deserts of Africa have very low densities of livestock. In arid and semiarid regions of Africa, where the potential for crop growth is limited, cattle, sheep, goats, and camels
are raised in low productivity, pastoral (extensive livestock
grazing) systems in which ambulatory stock can take advantage of seasonal, patchy vegetation growth. In these areas,
raising livestock is the only viable form of agriculture. In the
more settled humid, subhumid, and tropical highland areas,
cattle and small ruminants largely live in the same areas as
the human population. In these mixed crop-livestock farming
systems, livestock can increase crop production by providing draft power and manure, and by enhancing labor productivity. At the same time, organic material not suited for
human consumption can be converted into high-value food
and nonfood products, such as traction, manure, leather,
and bone.
livestock productivity must play an integral role in strategies

to reduce poverty and increase agricultural productivity.
Progress in poverty reduction will require well-targeted interventions to promote economic growth that the poor can
contribute to and from which they can benefit. Livestock
maps such as these, along with other information such as poverty and production systems, can contribute significantly to
better targeting.

The Gridded Livestock of the World database (Wint and
Robinson 2007) provided the first modelled livestock densities of the world, adjusted to match official national estimates for the reference year 2005 (FAO 2007), at a spatial
resolution of 3 arc-minutes (about 25 km2 at the equator). Recent methodological improvements have significantly enhanced these maps. More up-to-date and detailed
subnational livestock statistics have been collected; a new,
higher resolution set of predictor variables based on multi
temporal Moderate Resolution Imaging Spectroradiometer
(MODIS) imagery is used; and the analytical procedure has
been revised and extended to include a more systematic
assessment of the model accuracy. While the observed, subnational statistics vary in date and resolution, the maps are
standardized so that the national totals match the official
estimates for 2006 (FAO 2013).
Poverty in Africa remains widespread (p. 77). One quarter of

the worlds estimated 752 million poor livestock keepers live in
Africa south of the Sahara (SSA), where more than 85 percent
of them live in extreme poverty (Otte et al. 2012). Agricultural
productivity gains and diversification into high-value products such as livestock are essential ways of raising rural
incomes and improving food security in such areas. For three
reasonsthe large share of the rural poor who keep livestock,
the important contributions livestock can make to sustainable rural development, and the fast-growing demand for livestock productsdiversification into livestock and increased
Download the data from the Livestock-Geo-Wiki Project:

http://livestock.geo-wiki.org
26
Mapping the Global Distribution of Livestock.

Robinson et al. 2014.
The Food and Agriculture Organizations Gridded
Livestock of the World. Robinson, Franceschini, and
Wint 2007.
Gridded Livestock of the World, 2007.
Wint and Robinson 2007.
ATLAS
Ruminant livestock distribution, 2006

MAP 1Cattle
Number per km
MAP 2Sheep
Number per km
0
1
5
10
20
50
100
250
0
1
5
10
20
50
100
250
No data
No data
MAP 3Goats
Number per km
MAP 4Camels
Number per km
0
1
5
10
20
50
100
250
0
1
5
10
20
50
100
250
No data
No data
Data source
(all maps):
Robinson
et al. 2013;
Lecksell and
World Bank
2013.
27
Cropping Intensity
Stefan Siebert, Petra Dll, and Felix T. Portmann

The map shows cropping intensity, which is the number of
crop harvests per cell per year.1 Cropping intensity is highest
in irrigated regions, such as the Nile Delta (p. 19), or in wetland rice-growing areas, such as southern Nigeria and Cte
dIvoire, where more than one crop harvest per year is possible. In contrast, many rainfed areas in Africa see less than
one harvest per year due to scarce water or nutrient supplies,
particularly in drier regions such as the Sahel, South Sudan,
Central African Republic, and much of southern Africa.
Additionally, shifting cultivation, in which crops are grown
every three to ten years on available cropland with fallow
periods in between to allow for nutrient regeneration, is
common practice in Africa. These limitations and practices
lead to low cropping intensity values on average for most
regions of Africa (Figure 1). One also can use the map to
identify potential target areas for agricultural intensification
by identifying regions with low-cropping intensity and comparing them with areas with fast-growing populations.

The growing demand for agricultural products requires
either the cultivation of more land or intensified agricultural land use. It would be difficult to increase cropland
area, particularly in regions with high population density,
sensitive ecosystems, or poor soil quality. In such regions,
intensifying agricultural land use may be the only option.
Previous research on crop productivity has focused primarily on crop yields or yield gaps and therefore strictly on the
amount of crop yield per harvest. This works for temperate
climate regions where only one harvest is possible per year.
In contrast, in tropical or subtropical regions, increasing the
number of harvests per year can lead to increases in crop
production. Increasing cropping intensity by reducing the
length of the fallow period is a traditional way to adapt cultivation systems to growing demand for crop products and
to shortages in cultivatable land. To be sustainable, increases
in cropping intensity must be supplemented with improved
water and nutrient management.

Cropping intensity was calculated based on the
MIRCA2000 dataset (Portmann, Siebert, and Dll 2010) as
28
FIGURE 1
Cropping intensity by region, 2000
1.0
Average number of
crop harvests/year
0.8
0.6
0.4
0.2
0.0
Southern
Africa
Middle
Africa
Eastern
Africa
Northern
Africa
Western
Africa
Data source: Siebert, Portmann, and Dll 2010 and FAO 2012.
Note: Cropping intensity=the number of crop harvests per year.
a ratio of harvested crop area to cropland extent, which

included fallow land. This dataset provides, separately for
irrigated and rainfed agriculture, monthly growing areas of
26 crops or crop groups at a 5 arc-minute resolution. It refers
to the period around 2000 and was developed by combining global inventories on cropland extent (Ramankutty et al.
2008; Map 1, p. 17); the harvested area of 175 distinct crops
(Monfreda, Ramankutty, and Foley 2008); the extent of area
equipped for irrigation (Siebert, Hoogeveen, and Frenken
2006); and inventories on irrigated area per crop that used
crop calendars derived from FAO and other databases.

Current Opinion in Environmental Sustainability:
www.sciencedirect.com/science/journal/18773435/5/5
FAO Irrigated Crop Calendars: http://bit.ly/1c9yLH6
Global Estimation of Monthly Irrigated and
Rainfed Crop Areas: http://bit.ly/1dc6Gz8
Global Patterns of Cropland Use Intensity. Siebert,
Portmann, and Dll 2010: http://bit.ly/1rnJ0RT
Increasing Global Crop Harvest Frequency: Recent Trends
and Future Directions. Ray and Foley 2013:
http://bit.ly/1gTax8S
ATLAS
MAP 1
Cropping intensity
Average number of
harvests per year
0.0
0.2
0.4
0.6
0.9
1.1
1.3
1.5
3.0
No data
Data source: Siebert, Portmann, and Dll 2010.
29
Land Productivity for Staple Food Crops
Ulrike Wood-Sichra and Stanley Wood

Land productivity serves as a compact measure of the general status of agricultural and rural development. It is an
implicit reflection of the status of local environmental conditions, input use, and farmer know-how. Its spatial variation, furthermore, provides a picture of the likely relative
differences in land rental values. Detailed empirical studies of
diversity in land productivity point to a range of associated
factors including agroecology; farmers access to knowledge;
inputs, credit, infrastructure, and markets; land tenure; and
cultural preferences that shape crop and technology choices,
production practices, and market engagement.
Subtropicsemiarid
546
1448
501
1355
726
295
392
338
336
532
349
837
837
Subtropicsubhumid
Subtropichumid
Tropicarid
AFRICA
781
Western
Southern
Subtropicarid
Northern
Agroecological
zone
Middle
Almost three-quarters of Africas harvested agricultural land

is devoted to the production of staple food crops,1 but only
about one-third of that land generates annual output worth
more than $5002 from each cropped hectare. With farmers
typically cultivating just a half to three hectares of land to
support entire families, rural poverty and food insecurity are
pervasive, especially where nonfarm employment options are
limited. While some areas can produce food crop outputs
worth more than $2,500 per hectare (compared to an average of $517 per hectare across all of Africa), such impressive
results are concentrated in less than 1 percent of the total
harvested area and are likely boosted by access to irrigation.
Map 1 shows the distribution of Africas average land productivity for staple crops ranging from $250 or less per hectare at the fringes of the Sahel and in parts of eastern Africa
to $1,000 or more per hectare in southern Nigeria, parts of
Ghana, and along the Nile Valley and Delta. Summarizing
values by agroecological zone (p. 34), tropical arid zones,
such as on the northern edge of the Sahel and in eastern
Africa, have some of the lowest average values of production
per hectare; and subtropical arid zones, such as the Nile Delta
where irrigation is widely practiced, and subtropical humid
zones in southern Africa, have some of the highest average
values of production per hectare (Table 1).
TABLE 1 Average value of staple food crop production

(US$) per hectare by region in Africa
Eastern
89
208
186
336
225
184
Tropicsemiarid
336
469
100
351
246
270
Tropicsubhumid
496
479
161
491
1083
760
Tropichumid
659
661
133
1571
1144
749
Average
480
536
433
398
580
517
Data source: You et al. 2012; FAO 2012; Sebastian 2009.

Note: All local prices converted to international dollars at 20042006 average
purchasing power parity exchange rates.
and area harvested (hectares) for 14 of the most widely

grown food crops during the period 19992001 derived for
each 5 arc-minute grid cell3 across Africa using the Spatial
Production Allocation Model (SPAM) 2000 (You et al. 2012),
and (2) prices and national value of production for each
crop over the same period (FAO 2012). The total value of
food crop production (VoP) for any grid cell is calculated as
the sum of the VoPs for each crop, where VoP is a product of
crop price and production. Land productivity is derived by
dividing the total VoP of the 14 crops by the total harvested
area of those same crops for each grid cell.

More information on the SPAM model: http://mapspam.info
FAOSTAT database: http://faostat3.fao.org/home/index.html

Estimates of land productivity were derived using two core
data sources: (1) average annual production (metric tons)
1 Harvested areas and production values include the following staple food crops: maize, sorghum, millet, rice, wheat, barley, cassava, sweet potatoes and yams,
bananas and plantains, Irish potatoes, beans, groundnuts, soybeans, and other pulses.
2 All local prices converted to international dollars at 20042006 average purchasing power parity exchange rates.
30
ATLAS
MAP 1
Land productivity for staple food crops, 2000
Average value of production

in cropland area ($ per hectare)
0
250
500
1,000
2,500
Nonstaple food crop area
Data source: You et al. 2012 and FAO 2012.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
31
Works Cited
FARMING SYSTEMS OF AFRICA
Azzarri, C., S. Wood, G. Hyman, E. Barona, M. Bacou, and Z. Guo. 2012. Sub-national Poverty
Map for Sub-Saharan Africa at 2005 International Poverty Lines (r12.12).
http://bit.ly/1gq4jLF.
Dixon, J., J-M. Boffa, and D. Garrity. 2014. Farming Systems and Food Security
in Sub-Saharan Africa: Priorities for Science and Policy. Unpublished. Australian Centre
for International Agricultural Research and World Agroforestry: Canberra and Nairobi.
Dixon, J., A. Gulliver, and D. Gibbon. 2001. Farming Systems and Poverty: Improving Farmers
Livelihoods in a Changing World. Rome and Washington, DC: Food and Agriculture
Organization of the United Nations and World Bank. http://bit.ly/1dDekBW.
FAO (Food and Agriculture Organization of the United Nations). 2013a. Aquastat database.
Accessed on February 7, 2014. http://bit.ly/1gNqgEf.
. 2013b. FAOSTAT database. Accessed on Dec. 15, 2013. http://faostat3.fao.org/.
. 2013c. Global Livestock Production and Health Atlas database. Accessed on
February 7, 2014. http://bit.ly/NyPMF8.
. 2013d. Gridded Livestock of the World database. Accessed on February 7, 2014.
http://bit.ly/1pW6kFE.
. 2013e. State of the Worlds Land and Water Resources for Food and Agriculture
database. Accessed on February 7, 2014. www.fao.org/nr/solaw/en/.
FAO/IIASA (International Institute for Applied Systems Analysis). 2012. GAEZ
v3.0 Global Agro-ecological Zones database. Accessed on Feb. 7, 2014.
www.gaez.iiasa.ac.at/.
Garrity, D., J. Dixon, and J-M. Boffa. 2012. Understanding African Farming Systems: Science
and Policy Implications. Paper presented at the Food Security in Africa: Bridging
Research into Practice Conference, Australian International Food Security Centre/
Australian Centre for International Agriculture Research, Sydney, Australia, November
2012. http://bit.ly/1h8lmGJ.
HarvestChoice. 2011. Average Travel Time to Nearest Town Over 20K (hours) (2000).
Washington, DC and St. Paul, MN: International Food Policy Research Institute and
University of Minnesota. http://harvestchoice.org/node/5210.
UN (United Nations). 2013. World Urbanization Prospects, the 2011 Revision.
Department of Economic and Social Affairs, Population Division, Population Estimates
and Projections Section. Accessed February 7, 2014. http://esa.un.org/unup/.
CROPLAND AND PASTURELAND

Bartholom, E., and A. S. Belward. 2005. GLC2000: A New Approach to Global Land Cover
Mapping from Earth Observation Data. International Journal of Remote Sensing 26 (9):
19591977.
FAO. 2012. FAOSTAT database. Accessed October 2012.
. 2013. FAOSTAT: Concepts & Definitions: Glossary List.
Foley, J. A., R. DeFries, G. P. Asner, C. Barford, G. Bonan, S. R. Carpenter, F. S. Chapin, M. T.
Coe, G. C. Daily, H. K. Gibbs, J. H. Helkowski, T. Holloway, E. A. Howard, C. J. Kucharik,
C. Monfreda, J. A. Patz, I. C. Prentice, N. Ramankutty, and P. K. Snyder. 2005. Global
Consequences of Land Use. Science 309 (5734): 570574.
Friedl, M. A., D. Sulla-Menashe, B. Tan, A. Schneider, N. Ramankutty, and X. M. Huang. 2010.
MODIS Collection 5 Global Land Cover: Algorithm Refinements and Characterization
of New Datasets. Remote Sensing of Environment 114 (1): 168182.
Ramankutty, N., A. T. Evan, C. Monfreda, and J. A. Foley. 2008. Farming the Planet:
1. Geographic Distribution of Global Agricultural Lands in the Year 2000. Global
Biogeochemical Cycles 22 (1).
Thornton, P. K., and M. Herrero. 2010. Potential for Reduced Methane and Carbon Dioxide
Emissions from Livestock and Pasture Management in the Tropics. Proceedings of the
National Academy of Sciences of the United States of America 107 (46): 1966719672.
IRRIGATED AREAS
Siebert, S., P. Dll, J. Hoogeveen, J.-M. Faures, K. Frenken, and S. Feick. 2005. Development
and Validation of the Global Map of Irrigation Areas. Hydrology and Earth System
Sciences 9: 535547.
Siebert, S., V. Henrich, K. Frenken, and J. Burke. 2013a. Global Map of Irrigation Areas
Version 5. Bonn, Germany: Rheinische Friedrich-Wilhelms-University; Rome: Food and
Agriculture Organization of the United Nations. http://bit.ly/1eHpDex.
Siebert, S., V. Henrich, K. Frenken, and J. Burke. 2013b. Update of the Global Map of Irrigation
Areas to Version 5. Bonn, Germany: Rheinische Friedrich-Wilhelms-University; Rome:
Food and Agriculture Organization of the United Nations. http://bit.ly/1cM6bip.
CEREAL CROPS
FAO. 2012. FAOSTAT database. Accessed October 15, 2012.
32
You, L., Z. Guo, J. Koo, W. Ojo, K. Sebastian, M. T. Tenorio, S. Wood, and U. Wood-Sichra.
2012. Spatial Production Allocation Model (SPAM) 2000, Version 3, Release 1.
Accessed December 14, 2012. http://MapSPAM.info.
ROOT CROPS
LIVESTOCK AND MIXED CROP-LIVESTOCK SYSTEMS

Livelihoods in a Changing World. Rome: Food and Agriculture Organization of the
United Nations; Washington, DC: World Bank.
Robinson, T. P., P. K. Thornton, G. Franceschini, R. L. Kruska, F. Chiozza, A. Notenbaert, G.
Cecchi, M. Herrero, M. Epprecht, S. Fritz, L. You, G. Conchedda, and L. See. 2011. Global
Livestock Production Systems. Rome: Food and Agriculture Organization of the United
Nations and International Livestock Research Institute.
Ser, C., and H. Steinfeld. 1996. World Livestock Production Systems: Current Status, Issues
and Trends. FAO Animal Production and Health Paper 127. Rome: Food and Agriculture
Organization of the United Nations.
RUMINANT LIVESTOCK
FAO 2007. FAOSTAT database. Accessed in 2007. http://faostat.fao.org.
FAO. 2013. FAOSTAT database. Accessed in 2013. http://faostat.fao.org.
Lecksell, J., and World Bank. 2013. Personal communication regarding world boundaries for
2013, Nov. 26. Lecksell is lead World Bank cartographer.
Otte, J., A. Costales, J. Dijkman, U. Pica-Ciamarra, T. P. Robinson, V. Ahuja, C. Ly, and
D. Roland-Holst. 2012. Livestock Sector Development for Poverty Reduction: An
Economic and Policy PerspectiveLivestocks Many Virtues. Rome: Food and Agriculture
Organization of the United Nations, Animal Production and Health Division.
Robinson, T. P., G. Franceschini, and W. Wint. 2007. The Food and Agriculture
Organizations Gridded Livestock of the World. Veterinaria Italiana 43: 745751.
Robinson, T. P., G. R. W. Wint, G. Conchedda, T. P. van Boeckel, V. Ercoli, E. Palamara, G.
Cinardi, L. DAietti, S. I. Hay, and M. Gilbert. 2014. Mapping the Global Distribution of
Livestock. PLoS ONE, in press.
Wint, G. R. W., and T. P. Robinson. 2007. Gridded Livestock of the World, 2007. Rome:
Food and Agriculture Organization of the United Nations, Animal Production and
Health Division.
CROPPING INTENSITY
Monfreda, C., N. Ramankutty, and J. A. Foley. 2008. Farming the Planet: 2. Geographic
Distribution of Crop Areas, Yields, Physiological Types, and Net Primary Production in
the Year 2000. Global Biogeochemical Cycles 22 (1).
Portmann, F. T., S. Siebert, and P. Dll. 2010. MIRCA2000-Global Monthly Irrigated
and Rainfed Crop Areas around the Year 2000: A New High-Resolution Data Set for
Agricultural and Hydrological Modeling. Global Biogeochemical Cycles 24 (1).
Ramankutty, N., A. T. Evan, C. Monfreda, and J. A. Foley. 2008. Farming the Planet:
1. Geographic Distribution of Global Agricultural Lands in the Year 2000. Global
Biogeochemical Cycles 22 (1).
Ray, D. K., and J. A. Foley. 2013. Increasing Global Crop Harvest Frequency: Recent Trends
and Future Directions. Environmental Research Letters 8 (4): 110. http://bit.ly/1gTax8S.
Siebert, S., J. Hoogeveen, and K. Frenken. 2006. Irrigation in Africa, Europe and Latin
America: Update of the Digital Global Map of Irrigation Areas to Version 4. Frankfurt,
Germany: Goethe University; Rome: Food and Agriculture Organization of the United
Nations. http://bit.ly/1hm1f4Z.
Siebert, S., F. T. Portmann, and P. Dll. 2010. Global Patterns of Cropland Use Intensity.
Remote Sensing 2 (7): 16251643.
LAND PRODUCTIVITY FOR STAPLE FOOD CROPS

FAO. 2012. FAOSTAT database. Accessed October 15, 2012.
Sebastian, K. 2009. Agro-ecological Zones of Africa. International Food Policy Research
Institute. http://hdl.handle.net/1902.1/22616.
ATLAS
Growing Conditions
Agroecological Zones 34
Climate Zones for Crop Management 36
Rainfall and Rainfall Variability 38
Soil Fertility 40
Works Cited 42
33
GROWING CONDITIONS
Agroecological Zones
Kate Sebastian

Agroecological zones (AEZs) are geographical areas exhibiting similar climatic conditions that determine their ability
to support rainfed agriculture. At a regional scale, AEZs are
influenced by latitude, elevation, and temperature, as well
as seasonality, and rainfall amounts and distribution during
the growing season. The resulting AEZ classifications for
Africa have three dimensions: major climate (tropical or subtropical conditions), elevation (warmer lowland or cooler
upland production areas), and water availability (ranging
from arid zones with less than 70 growing days per year to
humid zones where moisture is usually sufficient to support crop growth for at least nine months per year) (Fischer
et al. 2009).
The map shows the broad latitudinal symmetry of
major climates and water availability north and south of
the equator, disrupted by the influence of highland and
lake complexes primarily associated with the East African
Rift Valley that extends from Ethiopia to Mozambique.
The Sahellocated between the Sahara Desert in the
north and the Sudanian Savanna in the southcomprises
warm tropical arid and semiarid zones characterized by a
strong north-south water availability gradient, while the
highlands of East Africa are distinguished by cooler, more
humid tropical conditions. The most extensive humid
zone is centered on the Congo Basin, stretching from the
Rwenzori and Virunga mountains at the borders of Uganda,
Rwanda, and the Democratic Republic of the Congo in the
east to the Atlantic coast in the west. The continent is primarily tropical, but significant subtropical areas with pronounced seasonality in temperatures and day length are
found in northern and southern Africa (beyond the tropical
limits of 23.44 degrees north and south of the equator).

Most African farmers, particularly in tropical areas, rely on
rainfed agriculture with very limited use of inputs such as
fertilizers. This means that the lands agricultural production depends almost solely on the agroecological context.
The spatial distribution of Africas dominant farming systems
(p. 15) is, therefore, closely aligned with the regional pattern of AEZs. Local agroecological conditions not only influence the range of feasible agricultural enterprise options but
also often strongly predict the feasibility and effectiveness
34
of improved technologies and production practices. For this

reason agriculture research and development planners are
keen to understand the nature and extent of agroecological
variation in the areas where they work. Planners who think
in terms of AEZ boundaries rather than country or regional
boundaries open up the potential for sharing knowledge and
tools with people on the opposite side of the continent who
work in similar AEZs. There is also growing interest in the
potential consequences of agroecological change. Change
might be brought about by mitigating local agroecological
constraints through, for example, investments in irrigation or
improved soil-water management practices. Or external factors such as climate change may drive agroecological change.
The likely negative economic and social implications of shifting agroecological patterns in Africa due to climate change
are priorities for emerging research and policy research.

The most common approaches to agroecological zone mapping were originally developed by the Food and Agriculture
Organization of the United Nations (FAO) and are still being
developed and applied (for example, Fischer et al. 2009 and
FAO/IIASA 2012). In Africa, the variable (and in many cases
declining) quality and availability of climate-station data
needed to generate reliable climatological maps is an
ongoing challenge (p. 37), although increased access
to satellite-derived weather and land-surface observations
could ease the constraints on gathering the data in the
future. The map was developed applying the regional AEZ
approach using long-term average, spatially interpolated climate data for Africa for the period 19601990 (Hijmans et al.
2005; Sebastian 2009).

AEZ maps and underlying data can be downloaded at:
http://hdl.handle.net/1902.1/22616
The most comprehensive collection of
global AEZ-related data can be found at the FAO/
International Institute for Applied Systems Analysis
Global Agro-Ecological Zones website:
www.fao.org/nr/gaez/en/
ATLAS
MAP 1
Agroecological zones
Tropic of Cancer
Su
bt
Su rop
bt ics
Tr rop - w
op ics ar
Tr ics - - co m
op w ol
ics arm
-c
oo
l
Equator
Arid
Semiarid
Subhumid
Humid
NA
Tropic of Capricorn
Data source: Sebastian 2009.

Note: Moisture classes are defined as follows: Arid=length of growing period (LGP) of less than
70 days; Semiarid=LGP of 70180 days; Subhumid=LGP of 180270 days; and Humid=LGP of
greater than 270 days.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
35
GROWING CONDITIONS
Climate Zones for Crop Management
Lieven Claessens and Justin Van Wart
Agricultural climate zones represent ecological conditions
farmers face based on moisture availability, length of growing period, and seasonality. Zones with little seasonal variation in temperature and in wet conditions are primarily
found in central Africa whereas the northernmost and
southernmost countries experience high temperature seasonality and arid conditions. This map provides information not only on what growing conditions these agricultural
climate zones present, but also on the relative size of each
zone. In some countries the climate zones are quite large,
such as in Mali or Niger where the weather is homogenous across large areas. In other countries, such as Kenya or
Ghana, these zones are much smaller as agricultural systems
face more diverse climates across space due to topography,
proximity to the coast, and/or rainfall variation. The relative size and extent of these zones offer information on the
expected diversity of cropping systems within each country and can be used to understand how effectively research
and technology can be extrapolated to other regions.
Table 1 provides a general understanding of the density and
average harvested area of zones within each region.

While agroecological zones (p. 34) help broadly define environments where specific agricultural systems may thrive,
an agriculture climate zone seeks to more adequately distinguish between the diversity of practices for similar agricultural systems within the larger agroecological zones,
primarily in terms of different climates. A map of agricultural
climate zones is a tool that can help scientists, governments,
and businesses determine the best areas to boost production or focus investment. These zones help streamline technology adoption and encourage innovative approaches by
providing insights into the size, location, and properties of
the climates where such technologies and research have
improved productivity. The map also helps identify similar
zones where new farming methods could be deployed in the
future to increase productivity of existing cropland. Knowing
the location of specific agricultural climate zones can help
stakeholders target new technologies and approaches to
the zones where they can make the most difference, and by
extension, help meet the growing demand for food in the
future. These agricultural climate zones can also be used to
scale up or extrapolate and compare site-specific results,
such as those obtained through field experiments or crop
36
TABLE 1 Agricultural climate zones and harvested area by

region of Africa
Number of
agricultural
climate zones
Average harvested
area per zone
(000 ha)
Northern Africa
72
446
Western Africa
39
2,425
Eastern Africa
71
853
Middle Africa
56
370
Southern Africa
77
80
126
680
Region
All Africa
Data source: van Wart et al. 2013 and FAO 2012.

Note: ha=hectares.
simulations, to larger regions or even other countries. For

example, new rice management systems being developed by
the AfricaRice organization for western Africa (Africa Rice
Center 2011) would also be useful in south central India and
central Thailand, where rice is grown in similar climate zones.

These observations are based on the Global Yield Gap Atlas
Extrapolation Domain (GYGA-ED) approach. The GYGA-ED
is constructed from three variables: (1) growing degree days
(GDD) with a base temperature of 0C; (2) temperature seasonality (quantified as the standard deviation of monthly
average temperatures); and (3) an aridity index (annual total
precipitation divided by annual total potential evapotranspiration). Each grid cell for weather data is approximately
100 km2 at the equator. Growing degree days and temperature seasonality were calculated using climate data from
WorldClim (Hijmans et al. 2005); the aridity index values
were taken from CGIAR-CSI (Trabucco et al. 2008) (p. 54). A
more extensive description and comparison with other zone
schemes can be found in van Wart et al. (2013).

Global Yield Gap Atlas: www.yieldgap.org
Zone characterizations: Use of Agro-Climatic Zones to
Upscale Simulated Crop Yield Potential.
van Wart et al. 2013.
Boosting Africas Rice Sector: http://bit.ly/1kBUWO3
ATLAS
MAP 1
Agricultural climate zones
High seasonality
Cooler
Wet
Arid
Low seasonality
Warmer
Data source: van Wart et al. 2013.

Note: The gradient on this map reflects three factors: moisture, temperature, and seasonality. Seasonality is based on variations in temperature and is quantified as the standard deviation of monthly average temperatures.
37
GROWING CONDITIONS
Rainfall and Rainfall Variability
Philip Thornton

In Africa, where most agriculture is rainfed, crop growth is
limited by water availability. Rainfall variability during a growing season generally translates into variability in crop production. While the seasonality of rainfall in the drier rangelands
can play a significant role in productivity, rain-use efficiency
(RUE)the amount of biomass produced (in kilograms of
dry matter per hectare) per millimeter of rainfallalso drives
production. RUE averages about 3.0 kg of dry matter per hectare for every millimeter of rainfall in northern Africa, 2.7 in
the Sahel, and 4.0 in eastern Africa, compared with up to
10.0 or so in temperate rangelands (Le Hourou, Bingham,
and Sherbek 1988). Estimates of annual rainfall variability
in the drier rangeland can offer a rough indication of possible production changes. Figure 1 shows how Ethiopias
gross domestic product echoed rainfall variability (measured as a percentage variation from the long-term average)
from the early 1980s to 2010. The close relationship illustrates the importance of rainfed agricultural production to
the national accounts of Ethiopia during this time period.
Ethiopia is one of many countries in Africa where the economy is closely tied to rainfed agriculture.
38
20
15
1.5
10
0.5
0
0.5
5
10
15
1.5
20
1982
1986
1990
Rainfall variability
1994
1998
Ag GDP growth
2002
2006
2
2010
Rainfall variability (WASP)
An average of less than 1,000 millimeters of rain falls per year

across most of Africa (Map 1). Rainfall tends to decrease
with distance from the equator and is negligible in the
Sahara (north of about latitude 16N), in eastern Somalia,
and in the southwest of the continent in Namibia and South
Africa. Rainfall is most abundant on the eastern seaboard
of Madagascar; portions of the highlands in eastern Africa;
large areas of the Congo Basin and central Africa; and parts
of coastal western Africa including Liberia, Sierra Leone, and
Guinea. Northern Africa experiences highly variable rainfall, except along the coasts of Algeria and Morocco (Map 2).
This regions coefficient of variationa measure of how
much rainfall varies from the annual averageis greater
than 45 percent, reflecting the erratic nature of rainfall in
a region that gets little precipitation. The story is similar in
the extreme southwest of the continent and in pockets of
the Horn of Africa. The amount of rainfall in parts of the
Congo Basin is much less variable, with a coefficient of variation around 1015 percent. For most of the continent where
rainfed crops are prevalent, the variability is 1535 percent.
FIGURE 1 Economic growth and rainfall variability in

Ethiopia, 19822010
Percentage change in GDP
GDP growth
Source: Thornton, Ericksen, and Herrero 2013; World Bank 2013; IRI/LDEO 2013.
Note: WASP = the 12-month Weighted Anomaly of Standardized Precipitation.

Rainfall data are from WorldClim (Hijmans et al. 2005),
an interpolated product based on average monthly climate data from weather stations from 1960 to 1990. The
data were aggregated to a spatial resolution of 5 arc-
minutes (grid cells approximately 100 km2 at the equator), and the long-term average monthly rainfall amounts
add up to the annual totals (Map 1). To estimate the
variability of annual rainfall (Map 2), the weather generator MarkSim (Jones and Thornton 2013) was used to
simulate 1,000 years of daily rainfall data for the roughly
420,000 grid cells that make up Africa and the standard
deviation of annual rainfall was calculated for each grid cell
and converted to the coefficient of variation. MarkSim predicts rainy days and is able to simulate the variation in rainfall observed in both tropical and temperate regions.

WorldClim data. Hijmans et al. 2005:
www.worldclim.org/methods
Generating Downscaled Weather Data from a Suite of
Climate Models for Agricultural Modelling Applications.
Jones and Thornton 2013.
Evidence from Rain-use Efficiencies Does Not Indicate
Extensive Sahelian Desertification.
Prince, Brown De Colstoun, and Kravitz 1998.
ATLAS
MAP 1
Average annual rainfall

10W
10E
20E
30E
40E
50E
Millimeters per year
0 or missing value
0<
250
500
1,000
1,500
2,000
2500
30N
20N
10N
MAP 2
Variability in annual rainfall

10S
20S
Percent
0
15
25
35
45
30S
Data source: Map 1WorldClim (Hijmans et al. 2005); Map 2MarkSim (Jones and Thornton 2013).
Note: Rainfall variability is represented by the coefficient of variability (CV), calculated as the
standard deviation divided by the mean annual rainfall. It is expressed as a percentage and indicates how much rainfall varies from average annual rainfall.
39
GROWING CONDITIONS
Soil Fertility
Cindy Cox and Jawoo Koo

Years of weathering have leached nutrients away from many
soils in the cropped areas of Africa south of the Sahara (SSA).
The resulting highly acidic soils (< 5.5 pH) are vulnerable to
aluminum toxicity, an issue across much of Africa (Map 1),
which occurs when aluminum becomes soluble and poisons
plants. It is the most common soil constraint across major
farming systems in SSA (Figure 1), affecting 32 percent of
cropland, followed by low nutrient reserves (20 percent) and
high leaching potential (12 percent). The worst soils in SSA
are concentrated along the eastern coast, throughout central
Africa, and scattered throughout the Sahel (Map 2). The Sahel
and central Africa suffer primarily from high-leaching potential and low-nutrient reserves. Some soils along eastern Africas
coastal edges and in the Horn of Africa are calcareous,
containing high levels of calcium carbonate. Such soils can
be highly fertile, but extremely calcareous soils can make
crops nutritionally deficient by fixing phosphorus (P), which
makes it insoluble and therefore not available to plants. SSA
is also home to large expanses of fertile soils that are free
of constraints.
decrease acidity in soils. Breeder selection for crop varieties, such as beans, sorghum, and fodder crops that resist
aluminum toxicity, is another way to deal with toxic soils.
Furthermore, the consequences of poor soil fertility can exacerbate other constraints, such as water uptake. Understanding
where and how soils are constrained is a primary concern for
the farmers and stakeholders who depend on less than ideal
soil conditions and those who seek to improve their welfare.

The underlying spatial data for major soil constraints was
taken from an updated version of the Soil Functional
Capacity Classification System (FCC) (HarvestChoice
2010). HarvestChoice updated the FCC by applying version 4s methodology (Sanchez, Palm, and Buol 2003)
to FAOs Harmonized World Soil Database version 1.1.
The pixel-level FCC data (Palm et al. 2007) was aggregated
using HarvestChoices cropland extent estimate (You et
al. 2012) and FAOs Farming Systems (Dixon, Gulliver, and
Gibbon 2001; p. 14).
About 80 percent of SSAs cropland is not considered highly

suitable for agriculture, because the extremely weathered soil
limits farmers yields. Low-input farming further degrades
soils when farmers fail to replenish nutrient reserves mined
by crops. To combat poor soil, liming can increase pH and
FIGURE 1
Updating Soil Functional Capacity Classification System,

HarvestChoice 2010: http://harvestchoice.org/node/1435
Africa Soil Information Service data: http://africasoils.net
Dominant soil constraint by farming system type in Africa south of the Sahara
Agropastoral
Maize mixed
Cereal-root crop mixed
Pastoral
Root and tuber crop
Humid lowland tree crop
Highland mixed
Highland perennial
Irrigated
Perennial mixed
Artisanal fishing
Aluminum toxicity
Calcareous
Cracking clays
High-leaching potential
High P fixation
Low nutrient reserves
Poor drainage
Volcanic
Free of constraints
0
10
12
14
Percent of total cropped area

Data source: Dixon, Gulliver, and Gibbon 2001; Sanchez, Palm, and Buol 2003; HarvestChoice 2010.
Note: See glossary for definitions of specific soil constraints.
40
16
18
20
22
ATLAS
MAP 1
Dominant soil constraints within cropped areas of Africa south of the Sahara (SSA)
Dominant soil constraint
Aluminum toxicity
Calcareous
Cracking clays
High-leaching potential
High P fixation
Low nutrient reserves
Poor drainage
Volcanic
Free of constraints
Outside crop growing area
Outside focus area
MAP 2 Area of cell affected by soil

constraints within cropped areas of SSA
Percent of grid cell

0
50
100
Outside crop growing area
Outside focus area
Data source: Map 1Sanchez, Palm, and Buol 2003; HarvestChoice 2010; You et al. 2012;
Map 2HarvestChoice 2010 and You et al. 2012.
Note: Grid cells are approximately 100 km2 at the equator. See glossary for definitions of
specific soil constraints.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
41
GROWING CONDITIONS
Works Cited
AGROECOLOGICAL ZONES
FAO (Food and Agriculture Organization of the United Nations) and IIASA (International
Institute for Applied Systems Analysis). 2012. Global Agro-ecological Zones.
www.fao.org/nr/gaez/en/.
Fischer, G., M. Shah, H. van Velthuizen, and F. Nachtergaele. 2009. Agro-ecological Zones
Assessments. In Land Use, Land Cover and Soil Sciences, Vol. III, edited by W. H. Verheye.
Oxford, UK: United Nations Educational, Scientific, and Cultural Organization/
Encyclopedia of Life Support Systems.
Hijmans, R. J., S. E. Cameron, J. L. Parra, P. G. Jones, and A. Jarvis. 2005. Very High
Resolution Interpolated Climate Surfaces for Global Land Areas. International Journal
of Climatology 25: 19651978.
CLIMATE ZONES FOR CROP MANAGEMENT

Africa Rice Center (AfricaRice). 2011. Boosting Africas Rice Sector: A Research for
Development Strategy 20112020. Accessed February. 12, 2014. http://bit.ly/1kBUWO3.
Trabucco, A., R. J. Zomer, D. A. Bossio, O. van Straaten, and L. V. Verchot. 2008. Climate
Change Mitigation through Afforestation/Reforestation: A Global Analysis of
Hydrologic Impacts with Four Case Studies. Agriculture, Ecosystems and Environment
126: 8197.
van Wart, J., L. G. J. van Bussel, J. Wolf, R. Licker, P. Grassini, A. Nelson, H. Boogaard, J.
Gerber, N. D. Mueller, L. Claessens, M. K. van Ittersum, and K. G. Cassman. 2013. Use
of Agro-Climatic Zones to Upscale Simulated Crop Yield Potential. Field Crops Research
143: 4455.
RAINFALL AND RAINFALL VARIABILITY

42
IRI/LDEO (International Research Institute for Climate and Society/Lamont-Doherty Earth

Observatory). 2013. Climate Data Library. http://iridl.ldeo.columbia.edu/index.html.
Jones, P. G., and P. K. Thornton. 2013. Generating Downscaled Weather Data from a
Suite of Climate Models for Agricultural Modelling Applications. Agricultural Systems
114: 15.
Le Hourou, H. N., R. L. Bingham, and W. Sherbek. 1988. Relationship between the
Variability of Primary Production and the Variability of Annual Precipitation in World
Arid Lands. Journal of Arid Environments 15: 118.
Prince, S. D., E. Brown De Colstoun, and L. L. Kravitz. 1998. Evidence from Rain-Use
Efficiencies Does Not Indicate Extensive Sahelian Desertification. Global Change Biology
4:359374. http://bit.ly/1g522Ug.
Thornton, P. K., P. J. Ericksen, and M. Herrero. 2013. Climate Variability and Vulnerability to
Climate Change: A Review. Global Change Biology, http://bit.ly/1pIHBBZ.
World Bank. 2013. World Development Indicators. http://data.worldbank.org/indicator.
SOIL FERTILITY
Livelihoods in a Changing World. Rome: Food and Agriculture Association of the United
Nations; Washington, DC: World Bank.
HarvestChoice. 2010. Updating Soil Functional Capacity Classification System.
HarvestChoice Labs. Accessed December 2, 2012. http://harvestchoice.org/node/1435.
Palm, C., P. Sanchez, S. Ahamed, and A. Awiti. 2007. Soils: A Contemporary
Perspective.Annual Review of Environment and Resources 32 (1): 99129.
Sanchez, P. A., C. A. Palm, and S. W. Buol. 2003. Fertility Capability Soil Classification:
A Tool to Help Assess Soil Quality in the Tropics. Geoderma 114: 157185.
http://bit.ly/NNLn0L.
ATLAS
Role of Water
Effects of Rainfall Variability on Maize Yields 44
Blue and Green Virtual Water Flows 46
Blue and Green Water Use by Irrigated Crops 48
Rainfall Data Comparison 50
Works Cited 52
43
ROLE OF WATER
Effects of Rainfall Variability on Maize Yields
Jawoo Koo and Cindy Cox
Farmers in SSA depend largely on rainfed crops for food security and their livelihood. But how reliable is rainfall across
Africa, and how does the variability of rainfall from season to
season affect crop yields? The following maps indicate where
the variability of total rainfall in SSA (Map 1) may influence
maize yields (Map 2), depending on the level of inputs such
as fertilizer used in maize cultivation (Map 3 and Map 4) and
the environment (Figure 1). A comparison of Maps 1 and
2 shows a correlation between rainfall and rainfed maize
yields. Yields tend to correspond to seasonal fluctuations in
rainfall, although in SSA, yields fluctuate year to year more
than rainfall since crops are at the mercy of many other factors, including total rainfall, cultural practices, pests, and soil
quality. Maps 3 and 4 show how the variability in yields may
be affected by changes in inputs. Figure 1 shows with more
inputs, such as hybrid seeds and more fertilizer (50 kilograms
of nitrogen per hectare), the probability of achieving acceptable levels of yield variabilityassumed to be 25 percent
or lessrises, although the effect of increased inputs, or
intensification, varies by agroecological zone (p. 34). When
shifting from low to high inputs, the share of total maize
growing area considered more reliablethat is, exhibiting
lower estimated variability in yieldrises from 20 percent
to 74 percent in the subhumid and humid regions of SSA.
In contrast, high inputs in arid and semi-arid regions of SSA
have a smaller impact on crop reliability with a change from
11 percent to 56 percent, as the yield potential in this region,
including the southern portions of Mali and Niger and central Chad, is more limited by water availability than in the
humid and subhumid regions of western Africa. In some
areas, such as the northern edge of the Sahel, the variability
may even rise (Map 4).

While estimates of yearly rainfall averages are important,
yield reliability, predicted by fluctuations in growing conditions from year to year, concerns farmers worldwide.
Knowing how rainfall variability affects yields helps stakeholders make climate-based decisions about what crops to
grow, which farming systems and management practices are
most suitable at a particular location, and where more investments and resources are needed to improve farm productivity and welfare. These may include decisions related to scaling
up technologies such as irrigation, synthetic fertilizers, hybrid
44
FIGURE 1 Variation in share of total maize growing area

under varying input levels by agroecological zone
Total maize growing area (%)
Agroecological zone
Arid and semiarid
100
90
80
70
60
50
40
30
20
10
0
Subhumid and humid
74
56
20
11
0
25
Input Level
Low
High
50
75
100 125 150 0
25
50
75
100 125 150
Variability of rainfed maize yield (%)

Data source: Elliott et al. 2014 and Sebastian 2009.
Note: The variability of rainfed maize yield is measured by the coefficient of
variation (CV).
maize, and improved crop varieties that are more resistant to

or better tolerate moisture fluctuations and drought.

Grid-based historical daily weather and soil databases were
used as inputs for the CERES-Maize model in the Decision
Support System for Agrotechnology Transfer (DSSAT)
v4.5 (Jones et al. 2003). Historical daily weather data for
19802010 generated by Elliott et al. (2014) based on
the AgMIP Hybrid Baseline Climate Dataset (Ruane and
Goldberg 2014) was used to retrieve site-specific solar radiation, temperature, and rainfall. The season-to-season variability in rainfall was measured using the coefficient of
variation (CV). The CV divides the standard deviation
by the mean, thus indicating the likelihood that rainfall in a given area will vary from the long-term average.
A gridded soil database was derived from FAOs Harmonized
World Soil Database v1.1 (FAO et al. 2009) and the ISRIC
WISE Global Soil Profile Database v1.1 (Batjes 2002).
The CERES-Maize model simulated rainfed maize production across the region in areas where rainfed maize production is biophysically possible. The modeling was performed
at a resolution of 5 arc-minutes, where a grid cell is approximately 100 km2 at the equator.

Rainfall Variability and Crop Yield Potential:
http://bit.ly/1jCMRbN
ATLAS
MAP 1 Variability of total rainfall during maize

growing season
Variability (CV)
Low
MAP 2
Variability of estimated maize yields
Variability (CV)
Low
High
High
Outside maize growing area
Outside focus area
Outside focus area
MAP 3 Variability of maize yield potential under

low inputs
Variability (CV)
Low
MAP 4Variability of maize yield potential under
high inputs
Variability (CV)
Low
High
High
Outside focus area
Outside focus area
Data sources: Map 1Ruane and Goldberg 2014; Elliott et al. 2013; Elliott et al. 2014;
Maps 24Authors using DSSAT model in Hoogenboom et al. 2011; Elliott et al. 2013; Elliott et al. 2014.
Notes: Rainfall variability based on seasonal total rainfall during maize growing period. Rainfed maize yield
variability estimated from simulated seasonal maize yield. Low inputs=open-pollinated seeds with no
fertilizer. High inputs=hybrid seeds with 50 kg nitrogen fertilizer per ha. All data simulated for 19501990.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
45
ROLE OF WATER
Blue and Green Virtual Water Flows
Stefan Siebert and Petra Dll

Production and consumption of agricultural commodities
used to be local. Now, with the rapid growth in trade and
urban areas, food may be produced in one place and consumed far away. With globalization, new links and dependencies between producers and consumers have formed.
Demand from faraway markets for agricultural commodities may elevate local resource use. On the other hand,
resource shortages in major production regions may result
in reduced crop yields and send price signals to commodity markets worldwide. Mapping virtual water flows helps
policymakers to better understand the importance of
links between resource use and trade and of dependencies
between producers and consumers of commodities.
1 Each cell measures approximately 100km2 or 10,000 hectares at the equator.
46
Net virtual water flows, 2000

40
30
20
1
10
0
-10
-1
-2
-3
-20
Blue virtual water

Green virtual water
Eastern
Africa
Middle
Africa
-30
Northern
Africa
Southern
Africa
Western
Africa
-40
Net flow of green virtual water (km per year)
The term virtual water content refers to the volume of water

used by a crop per unit of crop harvest. Virtual water flows
are then determined by commodity flows between the locations where crops are produced and consumed. Virtual
water flows are further distinguished as flows of blue (irrigation) and green (precipitation stored in the soil) water. The
maps show blue and green net virtual water flows caused
by the production and consumption of 19 major crops
(wheat, barley, rye, maize, rice, sorghum, millet, pulses, soybeans, groundnuts, sunflower, rapeseed, potatoes, cassava,
grapes, citrus, dates, cocoa, coffee). Negative values in the
maps indicate a net outflow of virtual water and show major
production areas where the amount of water used locally
to produce crops consumed elsewhere is greater than the
amount contained in crops consumed locally. Positive values indicate a net inflow of virtual water to major consumption areas.
The major irrigation regions are the source regions of
blue virtual water flows (blue in Map 1) while concentrations of rainfed crop production are the source of green virtual water flows (green in Map 2). Cities and other densely
populated regions represent the sinks of virtual water flows
(red in Maps 1 and 2). In total, northern and southern Africa
see a net inflow of both blue and green virtual water while
eastern, middle, and western Africa have a net inflow of blue
water but a net outflow of green water, indicating that crop
imports from irrigated production compensate for exported
rainfed crops (Figure 1).
FIGURE 1
Net flow of blue virtual water (km per year)
Data source: Hoff et al. 2014 and FAO 2012.

Note: Blue virtual water=irrigation water drawn from groundwater bodies
(aquifers) or surface water bodies (rivers, lakes, wetlands, or canals). Green virtual
water=precipitation stored in the soil and used by rainfed and irrigated crops.
Positive values represent net flows into each region.

Crop production, crop water use, and corresponding blue
and green virtual water content were computed by applying the Global Crop Water Model (Siebert and Dll 2010).
Crop consumption within each country was computed by
adding imports of the respective crop commodity to domestic crop production and then subtracting the corresponding
commodity exports derived from the Comtrade database
for the period 19982002 (UN 2009). It was assumed that
per capita commodity consumption is similar for all people
belonging to the same country. Production surpluses and
deficits within each country were leveled out by commodity flows (and linked virtual water flows) across increasingly
larger distances and finally the whole country, if required
(Hoff et al. 2014). The dataset refers to 19982002 and has a
spatial resolution of 5 arc-minutes.1

Water Footprint Network: www.waterfootprint.org/
Water Footprints of Cities: Indicators for Sustainable
Consumption and Production. Hoff et al. 2014:
http://bit.ly/1ogjdK1
ATLAS
MAP 1 Net virtual water flow of blue water

(irrigation), 2000

100
25
5
1
1
5
25
100
Outflow
Inflow
MAP 2 Net virtual water flow of green water

(precipitation stored in the soil), 2000

100
25
5
1
1
5
25
100
Outflow
Inflow
Data source (all maps): Hoff et al. 2014.

Note: Virtual water content refers to the volume of water used by the crop per
unit of crop harvest. Virtual water flows are then established by commodity
flows between the locations of crop production and crop consumption.
47
ROLE OF WATER
Blue and Green Water Use by Irrigated Crops
Stefan Siebert and Petra Dll

In these maps, blue water refers to irrigation water while
green water is precipitation stored in the soil that is also
used by irrigated crops. The values refer to the amount of
water that is evapotranspirated, or converted from soil water
to vapor and evaporated off plant stems and leaves. Blue and
green water use by irrigated crops is highest in regions with a
large extent of irrigated land (p. 18), high cropping intensity
(p. 28), and climate conditions causing a high evaporative
demand, for example, along the Nile River, in the northern
African countries of Morocco, Algeria, Tunisia, and Libya,
and in South Africa (Maps 1 and 2). The contribution of blue
water to total water use of irrigated crops (Map 3) depends
on the aridity of the site because irrigation is mainly used
to replace missing precipitation. The staple food crops with
the highest irrigation water use are rice (12.1 km3 per year),
wheat (11.1 km3 per year), and maize (9.0 km3 per year)
(Figure 1). Combined they account for a third of the total
blue water used for irrigation in Africa. More than 77 percent of the total irrigation water use is in northern Africa.

Although only 9 percent of the harvested crop area in Africa
is under irrigation, cereal production would decline by about
24 percent in Africa without the use of irrigation (Siebert
and Dll 2010). This highlights the importance of irrigation
for food security. On the other hand, irrigation accounts for
86 percent of global consumptive freshwater use (Dll et al.
2012) with contributions of more than 90 percent in many
African countries. Availability of freshwater therefore may
limit the use of irrigation in many regions. To identify regions
where expanding irrigation could increase future crop production, it is necessary to consider irrigated crops blue
water use along with freshwater availability (Bruisma 2009).
Green water use is also important to consider, because blue
and green water can be substituted for each other.
FIGURE 1 Blue water use by irrigated crop and region,

19982002
Fodder grasses
Others perennial
Others annual
Rapeseed
Soybeans
Coffee
Sunflower
Northern Africa
Barley
Southern Africa
Eastern Africa
Grapes
Sugar beets
Western Africa
Groundnuts
Middle Africa
Sorghum
Potatoes
Date palm
Pulses
Cotton
Citrus
Sugar cane
Maize
Wheat
Rice
0
10
12
14
km per year
Data source: Siebert and Dll 2010 and FAO 2012.
Note: Blue water use refers to the net irrigation water used by irrigated crops.
the period 19982002 and also represents multicropping.

By computing daily soil water balances, GCWM determines
evapotranspiration of blue and green water for each crop
and grid cell. GCWM assumes that crop evapotranspiration of irrigated crops is always at the potential level and not
restricted by water shortage. Water withdrawals for irrigation are higher than consumptive use because of losses and
water requirements for soil preparation and salt leaching.

Crop evapotranspiration was calculated by the Global Crop
Water Model (GCWM, Siebert and Dll 2008, 2010), distinguishing blue water use, or the evapotranspiration of irrigation water (also called consumptive irrigation water use)
from green water use (evapotranspiration of precipitation).
GCWM is based on the global land use dataset MIRCA2000
(Portmann, Siebert, and Dll 2010), which provides monthly
growing areas for 26 irrigated and rainfed crop classes for
48
16

FAO Aquastat: http://bit.ly/1dUQWqj
The Global Crop Water Model (GCWM): Documentation and
First Results for Irrigated Crops. Siebert and Dll 2008..
ATLAS
MAP 1Blue water use by irrigated crops, 2000

0
0<
2
10
20
50
200
500
1,700
Lakes
MAP 2Green water use by irrigated crops, 2000

0
0<
2
10
20
50
200
260
Lakes
MAP 3 Contribution of blue water to total water use of

irrigated crops
Percent
0
20
30
40
50
60
70
80
100
No irrigation
Lakes
Data source (all maps): Siebert and Dll 2010.

Note: Blue water use refers to the net irrigation water used by irrigated crops. Green water use
refers to precipitation water stored in the soil and used by irrigated crops.
49
ROLE OF WATER
Rainfall Data Comparison
Jawoo Koo and Cindy Cox
Regional rainfall data estimates for Africa can look significantly different depending on which data sources are used
and how the data are analyzed. Estimates rely on precipitation records from a variety of land-based weather station
networks with varying levels of data quality and spatio-
temporal coverage. For example, Maps 1 and 2 illustrate
average annual rainfall for 20002008 in Africa south of
the Sahara (SSA) at the same 0.5 spatial resolution, but
are derived from different data sources. Data from the
University of East Anglias Climate Research Unit Time Series
(CRU-TS) v3.10.01 (Map 1) shows less pixel-to-pixel variability than the University of Delawares Gridded Monthly Time
Series (GMTS) v2.01 data (Map 2). This suggests different
modeling algorithms and possibly the use of fewer observations. Map 3 shows the percentage difference between the
two, indicating where the rainfall estimation of GMTS is relatively higher (green) or lower (red) than CRU-TS. The differences are particularly evident in areas with low annual
rainfall such as the Sahel, since the significance of the difference between averages will be greater when average rainfall
values are low. Significant differences in rainfall estimations
in areas of southern Africa, particularly in Mozambique, also
exist. Compared with the CRU-TS dataset, GMTS calculates
a 17 percent higher average rainfall for the entire SSA region
(Figure 1).
FIGURE 1 Distribution of average annual total rainfall

from two climate data sources
4,000
Average annual total rainfall (mm)
3,000
2,000
1,180
1,000
680
180
1,362
797
233
0
CRU-TS v3.10.01
GMTS v2.01
Source: University of East Anglia 2013 and University of Delaware 2009.

Note: Each bar indicates a grid-cell-level value. The dotted black line indicates the
average across the SSA region, and the gray area shows +1/-1 standard deviation.
The y-axis precipitation totals are nine-year averages (20002008) at 0.5 grid cells.
methods. For this reason, researchers should not rely on

just one dataset. Depending on the research questions and
geographical areas of interest, the data source chosen may
introduce bias to the results. If possible, researchers should
compare data across multiple datasets to better understand
the range of uncertainties and to avoid reaching conclusions
that may inflate or understate the truth.
Gridded climate data allow researchers to compare variations in climate with other phenomena, such as crop yields
or areas suitable for crop growth. Variables other than precipitationincluding cloud cover, diurnal temperature
range, frost day frequency, daily mean temperature, and
monthly average daily maximum temperatureare also
available and can be used for similar comparisons. Rainfall
averages and patterns are important not only to African
farmers and stakeholders who rely on rainfed crops for food
security and livelihoods, but also to researchers and decisionmakers who need climate information to predict patterns
of agricultural productivity, effects of water management
technologies (such as drought-adapted crop varieties or
conservation agriculture), and potential changes in climate
projected over the coming decades. Climate-related datasets
from different sources are not identical because of limited
source dataperhaps because the network of weather stations is not dense enoughand differences in interpolation
Historic gridded climate databases from two sources, the

University of East Anglias CRU-TS v3.10.01 (2013) and
University of Delawares GMTS v2.01 (2009), were used in
the mapping and intercomparison analysis for the years
20002008. Both datasets are based on the same 0.5 degree
spatial resolution (~3,600 km2 at the equator). Annual rainfall data were computed for each grid cell, and their average
values across the years were mapped and compared with
each other.
50

University of East Anglia CRU climate data:
www.cru.uea.ac.uk
University of Delawares Gridded Monthly Time Series
(GMTS) v2.01 data: http://bit.ly/1m7HvHk
ATLAS
MAP 1 Average annual rainfall for 20002008 from the

University of East Anglia CRU-TS
MAP 2 Average annual rainfall for 20002008 from the

University of Delaware GMTS
mm per year
per grid cell
mm per year
per grid cell
Low: 0
Low: 0
High: 2,655
High: 3,911
Outside focus area
Outside focus area
MAP 3 Difference between CRU-TS and GMTS average

annual rainfall, based on CRU-TS
Percent
50
25
10
0
10
25
50
75
100
Outside focus area
Data sources: Map 1University of East Anglia 2013; Map 2University of Delaware 2009;
Map 3Calculation based on University of East Anglia 2013 and University of Delaware 2009.
Note: Each grid cell measures 0.5 degrees or ~3,600 km2 at the equator.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
51
ROLE OF WATER
Works Cited
EFFECTS OF RAINFALL VARIABILITY ON MAIZE YIELDS
Batjes, N. H. 2002. A Homogenized Soil Profile Data Set for Global and Regional
Environmental Research (WISE, Version 1.1). Report 2002/01. Wageningen, The
Netherlands: International Soil Reference and Information Centre. http://bit.ly/1gNrrDR.
Elliott, J., M. Glotter, N. Best, J. Chryssanthacopoulos, D. Kelly, M. Wilde, and I. Foster. 2014.
The Parallel System for Integrating Impact Models and Sectors (pSIMS). Prepared for a
special issue of Environmental Modeling & Software, forthcoming.
Elliott, J., D. Kelly, N. Best, M. Wilde, M. Glotter, and I. Foster. 2013. The Parallel
System for Integrating Impact Models and Sectors (pSIMS). In Proceedings of the
XSEDE13 Conference: Gateway to Discovery, chaired by N. Wilkins-Diehr, San Diego, CA,
July 2225. New York: Association for Computing Machinery.
FAO (Food and Agriculture Organization of the United Nations), IIASA (International
Institute for Applied Systems Analysis), ISRIC-World Soil Information, ISSCAS (Institute
of Soil Science-Chinese Academy of Sciences), and JRC (Joint Research Centre of the
European Commission). 2009. Harmonized World Soil Database (Version 1.1). Rome:
FAO; Laxenburg, Austria: IIASA. http://bit.ly/1ljrHQy.
Hoogenboom, G., J. W. Jones, P. W. Wilkens, C. H. Porter, K. J. Boote, L. A. Hunt, U. Singh, J. L.
Lizaso, J. W. White, O. Uryasev, F. S. Royce, R. Ogoshi, A. J. Gijsman, G. Y. Tsuji, and J. Koo.
2011. Decision Support System for Agrotechology Transfer (DSSAT) Version 4.5.1.023.
http://bit.ly/1i7FvgF.
Jones, J. W., G. Hoogenboom, C. H. Porter, K. J. Boote, W. D. Batchelor, L. A. Hunt, P. W.
Wilkens, U. Singh, A. J. Gijsman, and J. T. Ritchie. 2003. The DSSAT Cropping System
Model. European Journal of Agronomy 18: 235265.
Ruane, A. C., and R. Goldberg. 2014. AgMIP Hybrid Baseline Climate Datasets: Shifted
Reanalyses for Gap-filling and Historical Climate Series Estimation. Unpublished,
National Aeronautics and Space Administration, Washington, DC.
BLUE AND GREEN VIRTUAL WATER FLOWS

FAO (Food and Agriculture Organization of the United Nations). 2012. FAOSTAT Database.
Hoff, H., P. Dll, M. Fader, D. Gerten, S. Hauser, and S. Siebert. 2014. Water Footprints of
Cities: Indicators for Sustainable Consumption and Production. Hydrology and Earth
System Sciences 18: 213226. http://bit.ly/1ogjdK1.
52
Siebert, S., and P. Dll. 2010. Quantifying Blue and Green Virtual Water Contents in Global
Crop Production as Well as Potential Production Losses without Irrigation. Journal of
Hydrology 384 (34): 198217.
UN (United Nations). 2009. United Nations Commodity Trade Statistics Database.
Accessed on January 21, 2009. http://comtrade.un.org.
BLUE AND GREEN WATER USE BY IRRIGATED CROPS

Bruinsma, J. 2009. The Resource Outlook to 2050: By How Much Do Land, Water and Crop
Yields Need to Increase by 2050? Rome: Food and Agriculture Organization of the United
Nations. http://bit.ly/NNLJEy.
Dll, P., H. Hoffmann-Dobrev, F. T. Portmann, S. Siebert, A. Eicker, M. Rodell, G. Strassberg,
and B. R.Scanlon. 2012. Impact of Water Withdrawals from Groundwater and Surface
Water on Continental Water Storage Variations. Journal of Geodynamics 5960:
143156.
FAO (Food and Agriculture Organization of the United Nations). 2012. FAOSTAT Database.
Portmann, F. T., S. Siebert, and P. Dll. 2010. MIRCA2000-Global Monthly Irrigated
and Rainfed Crop Areas around the Year 2000: A New High-Resolution Data Set for
Agricultural and Hydrological Modeling. Global Biogeochemical Cycles 24 (1).
Siebert, S., and P. Dll. 2008. The Global Crop Water Model (GCWM): Documentation
and First Results for Irrigated Crops. Frankfurt, Germany: Goethe University.
http://bit.ly/1kOQHvS.
Siebert, S., and P. Dll. 2010. Quantifying Blue and Green Virtual Water Contents in Global
Crop Production as Well as Potential Production Losses Without Irrigation. Journal of
Hydrology 384 (34): 198217.
RAINFALL DATA COMPARISON

University of East Anglia Climatic Research Unit. 2013. CRU TS3.10: Climatic Research
Unit (CRU) Time-Series (TS) Version 3.10 of High Resolution Gridded Data of Monthby-month Variation in Climate (Jan. 1901Dec. 2009). www.cru.uea.ac.uk/.
University of Delaware, Center for Climatic Research, Department of Geography. 2009.
Terrestrial Precipitation: 19002008 Gridded Monthly Time Series (v2.01). Accessed on
September 13, 2011. http://bit.ly/1m7HvHk.
ATLAS
Drivers of Change
Influence of Aridity on Vegetation 54
Impacts of Climate Change on Length of Growing Period 56
Maize Yield Potential 58
Wheat Stem Rust Vulnerability 60
Benefits of Trypanosomosis Control in the Horn of Africa 62
Works Cited 64
53
DRIVERS OF CHANGE
Influence of Aridity on Vegetation
Antonio Trabucco and Robert Zomer

More than half of Africas population lives in arid, semiarid,
or dry subhumid areas. This means nearly 600 million people spread across 75 percent of the continents land area live
under ecological conditions where subsistence agriculture
may be only partially suitable. In such regions, people may find
it difficult to increase incomes from agriculture and improve
food security. In fact, there is a direct correlation between
aridity and prevailing vegetation and land use (Figure 1). While
humid conditions encourage plant growth, arid conditions
do not. One way plants adapt to the lack of rain is by limiting
their growth. Figure 1 shows the natural process where ecosystems evolve from bare land to herbaceous areas, shrub land,
and forests, as more humid conditions prevail. Land use, in
turn, also reflects human needs. In particular, agriculture follows specific patterns according to aridity. In semiarid areas
farmers rely mainly on rainfed subsistence agriculture, which
limits crop yields unless irrigation is adopted. In contrast,
highly productive agriculture systems are found in places with
more humid conditions, such as in southern Nigeria.

Because precipitation alone does not properly characterize vegetation water stresses across large regions, an aridity
54
100
Bare land
Cropland
Herbaceous
Shrub land
Forest
80
60
40
20
0
<
0.0 0.05
5
0.1 0.09
0
0.2 0.19
0
0.3 0.34
5
0.5 0.49
0
0.6 0.64
5
0.8 0.79
0
1.0 0.99
0
1.2
4
1.2
5
The aridity index measures the adequacy of the precipitation to satisfy vegetation water requirements. Large areas of
northern and southern Africa are dry with an aridity index
of less than 0.65. In contrast, central Africa is more humid,
with an aridity index that exceeds 0.65. Variations in dryness reflect Africas geography and topography. For example,
hyperarid zones, such as the Sahara and Namibia deserts,
which receive less than 100 mm of precipitation annually,
correspond to prevailing high pressure systems preventing
cloud formation over the western edges of subtropical areas.
Equatorial areas are more humid than other parts of Africa,
because low pressure systems and strong air convection condense the moisture into clouds, which lead to high precipitation. Dry northeast monsoon winds blowing in from the
Arabian Desert make eastern Africa less humid than other
equatorial regions, such as central Africa and the Gulf of
Guinea, to the west. Mountains, such as Mt. Kenya and Mt.
Kilimanjaro, block the passage of rain-producing weather
systems, creating more humid conditions in highland areas
and drier conditions on the shielded side of these highlands.
FIGURE 1Land cover types, by aridity
Area share by land cover (%)
Aridity index
Source: Trabucco and Zomer 2009.
Note: Aridity index=precipitation (mm)/potential evapotranspiration (PET mm).
index is calculated as the ratio of annual precipitation to

potential evapotranspiration (PET). Thus, the aridity index
measures how much rainfall is available to satisfy the water
demand of a type of vegetation. Using this formula, aridity index values increase with more humid conditions and
decrease with more arid conditions. Annual precipitation
was derived from the WorldClim database (Hijmans 2005).
PET was calculated using the Hargreaves method applied to
temperature parameter layers from the WorldClim database
and extraterrestrial radiation (Allen et al. 1998; Trabucco
and Zomer 2009). Although the aridity index map reflects
average conditions between 1950 and 2000, rainfall in arid
and semiarid regions is highly variable across space and time
(Map 2, p. 39). This variability relates to the randomness
of prevailing convective rains in arid regions, where short,
heavy storms can either hit or miss an area.

Global Aridity and PET (Potential Evapo-Transpiration)
Database: http://bit.ly/1hYD3Iv
The Climatology of Sub-Saharan Africa. Nicholson 1983.
Crop Evapotranspiration: Guidelines for Computing Crop
Water Requirements: http://bit.ly/1kCFdzq
Carbon Sequestration in Dryland Soil, Chapter 2 in
TheWorlds Drylands: http://bit.ly/13HBTpc
ATLAS
MAP 1
Aridity index
Tropic of Cancer
Equator
Aridity Index
hyperarid
arid
semiarid
dry subhumid
humid
0
0.05
0.10
0.20
0.35
0.50
0.65
0.80
1.00
1.25
Tropic of Capricorn
Data source: Trabucco and Zomer 2009.

Note: Aridity Index=precipitation (mm)/potential evapotranspiration (PET mm). The aridity index classes are
based on United Nations Environment Programme classifications (UNEP 1997).
55
DRIVERS OF CHANGE
Impacts of Climate Change on Length of Growing Period
Philip Thornton

Projections show that climate changes between now and
the 2050s may significantly affect the length of growing
periods (LGP) in Africa. LGP, expressed as number of days
per year, is a metric that integrates rainfall, temperature, and
some soil conditions to determine when crops grow in certain areas (Map 1). It is a useful proxy for season type in
the water-limited conditions that prevail in many parts of
the tropics. LGP ignores intervening drought periods and so
it is not always a good indicator of cropping success, but it
is often highly correlated with yields. Map 2 shows the projected percentage change in LGP in the 2050s compared
with current conditions, using a scenario of high greenhouse
gas emissions and several global climate models. Most of the
continent will see reductions in LGP, some of them severe.
Parts of eastern Africa, particularly the Horn of Africa, may
see some increases, but in these areas, current LGP is low
(90 days or less, Map 1). The climate models used to project LGP do not all agree on how the climate may change by
2050. Map 3 shows the variability in projections for LGP estimated from several climate models. Since areas with lower
values, such as much of central Africa, show more agreement between the various climate models, one can have
more confidence in projected LGP changes in these areas.
In areas with higher values, the climate models agree less,
meaning the projections of LGP change are less reliable.

To effectively adapt to climate change, farmers, governments,
and other stakeholders must understand the potential effects
on crop and livestock production. A contracted growing season can impact crop and livestock productivity, particularly in
areas where growing seasons are already short. Temperature
increases and rainfall changes could push some of these areas
to a point where cropping may fail in most years. Some farmers may be able to adapt to shorter growing seasons by planting varieties that mature more quickly; other farmers may
need to change to more drought- and heat-tolerant crops.
Increase in LGP may present more growing opportunities, but
it is uncertain how the change in growing time would impact
soil moisture. As climate changes, the distribution of crop
pests and diseases may change, too. Of course, LGP is only one
metric; the information shown here can be combined with or
compared to other aspects of projected climate changesuch
as temperature changesto create a more detailed picture of
how climatic shifts could affect crop growth and development.
56
TABLE 1 AtmosphereOcean General Circulation Models

used to estimate LGP changes to the 2050s
Model Name
(Vintage)
Institution
Resolution
(degrees)
BCCR_BCM2.0
(2005)
Bjerknes Centre for Climate Research
1.9 1.9
CNRM-CM3
(2004)
Mto-France/Centre National de
Recherches Mtorologiques, France
1.9 1.9
CSIRO-Mk3_5
(2005)
Commonwealth Scientific and

Industrial Research Organisation
Atmospheric Research
1.9 1.9
ECHam5 (2005)
Max Planck Institute for Meteorology
1.9 1.9
INM-CM3_0 (2004) Institute Numerical Mathematics
4.0 5.0
MIROC3.2
(medres) (2004)
Center for Climate System Research,

National Institute for Environmental
Studies, and Frontier Research Center
for Global Change
2.8 2.8
Ensemble average
Average climatology of the above

models
Source: For model details, see Randall et al. 2007.

The data are from downscaled climate projections. Because
differences between climate models may be quite large, particularly for projected changes in rainfall patterns and quantities, the means of six climate models (Table 1) form the
basis for generating daily weather data sequences plausible
for future climatologies. Jones and Thornton (2013) provide
details of the models used and the methods applied. LGP
is calculated daily using a water balance model that calculates available soil water, runoff, water deficiency, and the
ratio of actual to potential evapotranspiration (Ea/Et). The
growing period begins with 5 consecutive growing days and
ends with 12 consecutive nongrowing days; a growing day
has an average air temperature greater than 6C and Ea/Et
exceeding 0.35.

Methods used to develop this data and create these maps:
www.ccafs-climate.org/pattern_scaling/
More information on the effects of climate change:
Easterling et al. 2007.
Details on models used and methods applied: Jones and
Thornton 2013; and Jones, Thornton, and Heinke 2009.
ATLAS
MAP 1 Average length of growing period (LGP) for

current conditions, 2000s
Number of days per year

0 or missing value
1
30
60
90
120
150
180
210
240
MAP 2 Projected mean change in length of growing

period (LGP) in 2050
Percent
75
20
5
0
5
MAP 3 Variability among length of growing period (LGP)

projected values for 2050
Percent
0 or missing value
1
30
60
90
120
Data source (all maps): Jones et al. 2009.

Note: LGP variability is represented by the coefficient of variation (CV), calculated as the standard deviation
divided by the average LGP, expressed as a percentage.
57
DRIVERS OF CHANGE
Maize Yield Potential
Jawoo Koo
Map 1 portrays the broad spatial distribution of farm-level

rainfed maize production in Africa south of the Sahara.
While South Africa, the regions largest producer, consistently achieves national average yields in excess of 4 tons per
hectare (t/ha), the best performing of the remaining
countries, including major producers such as Ethiopia,
Malawi, and Zambia, typically average only around 2 t/ha.
Farmers in other large-producer nations, notably Nigeria,
Tanzania, and Kenya, have lower yields, around the regional
norms of 1.31.7 t/ha. Map 2 shows potential rainfed maize
yield, or the modeled patterns of achievable yields if key
yield constraints, in this case soil nutrient deficiencies, could
be overcome. If concerted development efforts helped
to achieve this goal, approximately 55 percent of the current maize production area could attain yields in excess of
3 t/ha, a threshold that signals the basic subsistence cereal
needs of smallholder families can likely be met, assuming
typical farm holdings and family size (UN Millennium
Project 2005). The gap between actual and potential
yields tends to vary systematically by production environment (Figure 1). In drier regions (those areas with less than
500 millimeters of rainfall per year, such as the Sahel),
estimated yield gaps are relatively modest, because the lack
of rainfall remains a key limiting factor to increased yields
even if soil fertility is improved.

The maize yield analysis and mapping shown here helps target and prioritize specific geographic areas where researchers and farmers can work to overcome common sets of
production constraints to enhance local livelihoods and
food security. Yield potential in many areas is much higher
than what farmers now achieve. In areas with higher levels
of rainfall, improving soil quality can provide much bigger
payoffs for farmers. However, reducing soil nutrient deficiencies is not a cure-all; other challenges, such as the increasing prevalence of pests and weed competition need to be
addressed. Also, some production areas are inherently less
suited to maize production using existing technologies and
practices. Particularly in drier areas, farmers may already
be achieving the potential that current varieties can support without further investments in small-scale irrigation or
more drought-tolerant varieties.
58
FIGURE 1 Actual (2000) vs. potential maize yields, Africa

south of the Sahara
Maize yield (tons per hectare)
10
8
Actual yield
Potential yield
6
4
2
0
<500
(7%)
5001000
(49%)
10011500
(37%)
15012000
(6%)
>2000
(1%)
Average rainfall: millimeters/year (Percent of maize area)

Data source: Actual yieldYou et al. 2012; potential yieldauthors calculations.
Note: Bars indicate the average yield in each annual rainfall category weighted
with maize harvest area, and the error bars indicate one standard deviation. Percentages in parentheses indicate the approximate share of maize production area
in each rainfall category.

Historical daily weather and soil databases generated by
HarvestChoice were used as inputs to the DSSAT v4.5 CERES
-Maize model (Jones et al. 2003; Hoogenboom et al. 2012) in
order to simulate yields across a 5 arc-minute grid (with
~100km2 grid cells at the equator) covering Africa. Historical
monthly rainfall data for 195090 were extracted from the
University of East Anglia CRU-TS v3.10 database (UEA 2011)
and temporally downscaled to daily weather by applying
satellite-observed daily rainfall patterns for 19972008
retrieved from the NASA-POWER Agroclimatic Database.
Gridded soil texture classes (sandy, loamy, and clayey) were
extracted from the updated Soil Functional Capacity
Classification (FCC) System (HarvestChoice 2010a).
Achieved yields were extracted from the SPAM database
(You et al. 2012).

Spatial Production Allocation Model: http://mapspam.info
Synthesized 100-Year Weather Data. HarvestChoice 2010b:
http://harvestchoice.org/node/1441
Updating Soil Functional Capacity Classification Systems
HarvestChoice 2010: http://harvestchoice.org/node/1435
ATLAS
MAP 1
Actual rainfed maize yield, c. 2000
MAP 2
Potential rainfed maize yield
Low (0 t/ha)
Low (0 t/ha)
High (16 t/ha)
High (16 t/ha)
Outside actual growing area
Outside potential growing area
Outside focus area
Outside focus area
Data sources: Map 1You et al. 2012; Map 2Author.

Note: t/ha=tons per hectare.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
59
DRIVERS OF CHANGE
Wheat Stem Rust Vulnerability
Yuan Chai and Jason Beddow

Much of the wheat-growing area of Africa is susceptible to
stem rust, a fungal disease of wheat. The map shows where
crops might be vulnerable to stem rust infection. Almost all
African areas where wheat production is relatively concentrated are vulnerable to the disease, including the northern growing areas in Morocco, Tunisia, Algeria, and the Nile
Valley, along with major growing areas in central Ethiopia,
southern Kenya, and South Africa. The map shows the diseases potential to pose a problem if wheat were grown
throughout the continent, although it is not grown in all of
the colored areas. In a typical year, the pathogen can persist year-round in the red areas, infecting wheat, rye, and barley. The climate of the blue areas is also hospitable to the
pathogen, but it cannot survive the entire year in those locations, usually because they become too hot, cold, or dry. For
infection to occur in these areas, the pathogen must be transported (primarily by wind) to the area each year.

Stem rust negatively affects food security by limiting wheat
production, which increases food prices. Though
much of the wheat grown worldwide is somewhat resistant to the disease, most of the older cultivars used by
many low-input farmers in Africa and elsewhere offer little resistance. Further, most of the worlds wheat varieties
have little resistance to new strains of the stem rust pathogen, collectively known as Ug99, that were first discovered in Uganda in 1998. These new strains could severely
shrink global wheat supplies. From its emergence in Uganda,
Ug99 has spread to infect wheat crops grown in other
African countries, including major wheat-producing countries such as Kenya, Ethiopia, and South Africa.
On average, Africas wheat-growing areas are highly
susceptible to stem rust compared with global norms
(Table 1). Based on these estimates along with the cereal
crop distributions (p. 20), about 64 percent of the worlds
wheat area, representing 71 percent of global wheat
output, is climatically vulnerable to stem rust infection, and the disease can persist year-round in about
13 percent of that area. By contrast, 90 percent of
Africas wheat-growing area, representing 87 percent of
its wheat output, is susceptible to stem rust, and the disease can persist year-round in about 71 percent of the continents wheat-growing area, representing 67 percent of
Africas wheat output. Thus, not only is Africas wheat crop
60
TABLE 1 Stem rust vulnerability and persistence in Africa

and major wheat-growing areas of the world
Vulnerable to
stem rust
Region
Persistent
year-round
Area (%) Output (%) Area (%) Output (%)
China
91.6
90.6
6.6
3.9
India
60.6
63.0
2.8
1.2
United States
53.5
56.6
0.7
1.1
Russia
22.8
21.9
0.0
0.0
Africa
90.0
86.9
70.6
66.6
Global
63.8
71.2
12.6
9.4
Data source: Calculated based on Beddow et al. 2013b and You et al. 2012.
Note: The percentages show the portions of the wheat harvested area (area %)
and wheat produced (output %) that are susceptible to stem rust infection. Authors calculations based on stem rust potential and harvested area and annual
production for wheat. The climate in vulnerable areas allows the pathogen to
infect a host during the growing season. Persistent year-round=areas where the
pathogen can become established and survive year-round.
more vulnerable to stem rust infection, the disease is more

likely to be present every year.

Global estimates of climatic suitability were derived
by modeling the response of the stem rust pathogen,
Puccinia graminis, to climatic factors such as soil moisture
and temperature as described by Beddow et al. (2013a).
For each 10 arc-minute pixel (~344 km2 at the equator)
globally, the model was used to estimate the relative climatic suitability for the pathogen to infect a crop host
during the growing season (vulnerability) and to survive year-round (persistence).

Puccinia graminis. Beddow et al. 2013a.
Measuring the Global Occurrence and Probabilistic
Consequences of Wheat Stem Rust. Beddow et al. 2013b.
Potential Global Pest Distributions Using Climex:
HarvestChoice Applications. Beddow et al. 2010.
Right-Sizing Stem-Rust Research. Pardey et al. 2013.
Tracking the movement of Ug99CIMMYT Rust Tracker:
http://rusttracker.cimmyt.org
ATLAS
MAP 1
Areas vulnerable to wheat stem rust
Seasonally vulnerable
Persistently vulnerable
Data source: Beddow et al. 2013b.

Notes: Seasonally vulnerable=areas in which the pathogen can grow during the
favorable season but cannot survive year-round. Persistently vulnerable=areas where
the pathogen can become established and survive year-round.
61
DRIVERS OF CHANGE
Benefits of Trypanosomosis Control in the Horn of Africa
Timothy Robinson, Giuliano Cecchi, William Wint, Raffaele Mattioli, and Alexandra Shaw
Using the Horn of Africa as an example, the maps illustrate
different steps in a methodology developed to estimate
and map the economic benefits to livestock keepers of controlling a disease (Shaw et al. 2014). Cattle are first assigned
to different production systems as shown in Map 1, illustrating for example, where mixed farming is heavily dependent on the use of draft oxen in Ethiopia, areas of Sudan and
South Sudan where oxen use is much lower, and the strictly
pastoral areas of Somalia and Kenya. Information on the
location of cattle and production systems is combined with
the distribution of tsetse fly species in the area (Map 2) to
estimate the presence and absence of trypanosomosis, a parasitic disease transmitted by the tsetse fly. Herd growth and
spread is modelled for the current situation, and for the simulated removal of trypanosomosis. The outputs of the model
are then presented as a map of the financial benefits to livestock keepers that would be realized from trypanosomosis
removal, expressed as US$ per km2 (Map 3). The estimated
total maximum benefit to livestock keepers, interpreted also
as the maximum level of losses avoided, in the Horn of Africa
amounts to nearly $2.5 billion, discounted at 10 percent
over 20 years to account for the opportunity cost of funds
an average of approximately $3,300 per square kilometer
of tsetse-infested area (Table 1). Map 3 shows how these
benefits vary spatially.

African animal trypanosomosis reduces the productivity of
livestock, especially cattle, when it sickens or kills them. It
also affects rural development and livelihoods more generally by limiting options for mixed farming and hindering a
balanced use of natural resources. Moreover, in many areas
the parasite causes sleeping sickness in people; a highly
debilitating disease which if not treated is lethal. Deciding
where and how to intervene against this disease requires
knowledge of relevant socioeconomic dimensions, such as
poverty levels (p. 76) and the role of livestock in peoples
livelihoods. The map of potential benefits from trypanosomosis removal in the Horn of Africa can help decisionmakers prioritize interventions by highlighting areas, such as
Ethiopia, South Sudan and Kenya, where the financial return
on investments to control the disease would be highest
(Table 1).
62
TABLE 1 Projected maximum benefits (US$) over 20 years

of eliminating bovine trypanosomosis
Total benefit
from absence of
trypanosomosis
(US$ million)
Average
benefit per
km2 infested
(US$)
Country
Area of tsetse
infestation
(000 km2)
Ethiopia
157
834
5,317
Kenya
129
590
4,576
38
158
4,181
South Sudan
and Sudan
310
485
1,564
Uganda
103
390
3,786
Total
737
2,457
3,335
Somalia
Source: Shaw et al. 2014.

Note: The total benefit represents the cumulative amount of money accrued
over 20 years, discounted at 10 percent to account for the opportunity cost of
funds, if trypanosomosis were removed in these countries over a five-year period.

The model used information on cattle densities and production systems to account for herd growth and spatial spread of
cattle over a 20-year period. For this analysis, pastoral, agropastoral, and mixed farming systems, as described in Cecchi
et al. (2010), were further characterized to measure dairy and
draft power in the Horn of Africa, using reported statistics on
improved cattle that were cross-bred with higher yield varieties and on the use of draft oxen. The cattle distribution map
used for the analysis was an earlier version of that presented
for the whole of Africa (Map 1, p. 27). The predicted presence
of six tsetse fly species of veterinary importance in eastern
Africa at one kilometer resolution (Wint 2001) were combined into a single regional map that predicts the absence or
presence of the genus Glossina (tsetse fly). Shaw et al. (2014)
describe the herd model used and the detailed data on herd
parameters with and without trypanosomosis in the region.

Mapping the Economic Benefits to Livestock Keepers from
Intervening Against Bovine Trypanosomosis in Eastern
Africa. Shaw et al. 2014.
Geographic Distribution and Environmental
Characterization of Livestock Production Systems in Eastern
Africa. Cecchi et al. 2010.
ATLAS
MAP 1
Cattle production systems

30E
MAP 2
Tsetse fly distribution

30E
45E
45E
Eritrea
Eritrea
15N
15N
Sudan
Djibouti
Sudan
Djibouti
Somalia
Somalia
South Sudan
South Sudan
Ethiopia
Ethiopia
Uganda
Uganda
Kenya
Kenya
! !
1,000
500
MAP 3
!
!
Predicted distributon of tsetse flies
1,000
Low oxen
Medium oxen
High oxen
Areas unsuitable for ruminants
Pastoral
500
Kilometers
Low oxen
Medium oxen
High oxen
High dairy
Low oxen
Medium oxen
High oxen
High dairy
Mixed farming (Ethiopia)
Mixed farming (general)
Agropastoral
Kilometers
Absence
Presence
Outside study area
Outside study area
Potential benefits of eliminating bovine trypanosomosis

45E
30E
Djibouti
Sudan
South Sudan
Ethiopia
US$ per km over

a 20-year period
0
10
500
Somalia
Uganda
1,000
2,500
Kenya
5,000
7,500
10,000
12,500
250
500
Unsuitable for ruminants

Outside study area
Kilometers
Data source: Maps 1 and 2Shaw et al. 2014; Map 3Calculation

based on Wint 2001.
Note: Map 3The benefit=total amount of money accrued over 20
years, discounted at 10 percent to account for the opportunity cost of
funds if trypanosomosis were eliminated in a five-year period.
63
DRIVERS OF CHANGE
Works Cited
INFLUENCE OF ARIDITY ON VEGETATION
Allen, R. G., L. S. Pereira, D. Raes, and M. Smith. 1998. Crop EvapotranspirationGuidelines
for Computing Crop Water Requirements. FAO Irrigation and Drainage Paper 56. Rome:
Food and Agriculture Organization of the United Nations.
Hijmans, R. J., S. E. Cameron, J. L. Parra, P. G. Jones, and A. Jarvis. 2005. Very High Resolution
Interpolated Climate Surfaces for Global Land Areas. International Journal of
Climatology 25: 19651978.
Nicholson, S. 1983. The Climatology of Sub-Saharan Africa. In Environmental Change in
the West African Sahel. Advisory Committee on the Sahel, 7192. Washington, DC:
National Academy Press.
Trabucco, A., and R. J. Zomer. 2009.Global Aridity and PET (Potential Evapotranspiration)
Database. CGIAR Consortium for Spatial Information. Accessed on September 25, 2013.
http://bit.ly/1hYD3Iv.
UNEP (United Nations Environment Programme). 1997. World Atlas of Desertication, 2nd
ed. London: UNEP.
IMPACTS OF CLIMATE CHANGE ON

LENGTH OF GROWING PERIOD
Easterling, W. E., P. K. Aggarwal, P. Batima, K. M. Brander, L. Erda, S. M. Howden, A. Kirilenko,
J. Morton, J.-F. Soussana, J. Schmidhuber, and F. N. Tubiello. 2007. Food, Fibre and Forest
Products. In Climate Change 2007: Impacts, Adaptation and Vulnerability: Contribution
of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on
Climate Change, edited by M. L. Parry, O. F. Canziani, J. P. Palutikof, P. J. van der Linden,
and C. E. Hanson, 273-313. Cambridge, UK, and New York: Cambridge University Press.
Jones, P. G., and P. K. Thornton. 2013. Generating Downscaled Weather Data from a Suite
of Climate Models for Agricultural Modelling Applications. Agricultural Systems 114:
15.
Jones, P. G., P. K. Thornton, and J. Heinke. 2009. Generating Characteristic Daily Weather
Data Using Downscaled Climate Model Data from the IPCCs Fourth Assessment. Project
Report. Nairobi: International Livestock Research Institute.
Randall, D. A., R. A. Wood, S. Bony, R. Colman, T. Fichefet, J. Fyfe, V. Kattsov, A. Pitman,
J. Shukla, J. Srinivasan, R. J. Stouffer, A. Sumi, and K. E. Taylor. 2007. Climate Models
and Their Evaluation. In Climate Change 2007: The Physical Science Basis: Contribution
of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel
on Climate Change, edited by S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis,
K. B. Averyt, M. Tignor, and H. L. Mille. Cambridge, UK, and New York: Cambridge
University Press.
MAIZE YIELD POTENTIAL

HarvestChoice. 2010a. Updating Soil Functional Capacity Classification System.
HarvestChoice Labs. Accessed December 2, 2012. http://harvestchoice.org/node/1435.
. 2010b. SLATE: Synthesized 100-Year Weather Data for Sub-Saharan Africa.
Washington, DC: International Food Policy Research Institute; St. Paul, MN: University of
Minnesota. http://harvestchoice.org/node/1441.
Hoogenboom, G., J. W. Jones, P. W. Wilkens, C. H. Porter, K. J. Boote, L. A. Hunt, U. Singh,
J. L. Lizaso, J. W. White, O. Uryasev, F. S. Royce, R. Ogoshi, A. J. Gijsman, G. Y. Tsuji, and
J. Koo. 2012. Decision Support System for Agrotechnology Transfer (DSSAT) Version
4.5.1.023. Honolulu: University of Hawaii. CD-ROM.
64
Jones, J. W., G. Hoogenboom, C. H. Porter, K. J. Boote, W. D. Batchelor, L. A. Hunt,

P. W. Wilkens, U. Singh, A. J. Gijsman, and J. T. Ritchie. 2003. The DSSAT Cropping
System Model. European Journal of Agronomy 18: 235265.
UN Millennium Project. 2005. Investing in Development: A Practical Plan to Achieve the
Millennium Development Goals. London and Sterling, VA, US: Earthscan.
University of East Anglia Climatic Research Unit. 2011. CRU TS3.10: Climatic Research
Unit (CRU) Time-Series (TS) Version 3.10 of High Resolution Gridded Data of Monthby-month Variation in Climate (Jan. 1901Dec. 2009). Accessed on September 13, 2011.
http://badc.nerc.ac.uk/home/index.html.
WHEAT STEM RUST VULNERABILITY

Beddow, J. M., T. M. Hurley, D. J. Kriticos, and P. G. Pardey. 2013b. Measuring the Global
Occurrence and Probabilistic Consequences of Wheat Stem Rust. Washington, DC and
St. Paul, MN, US: HarvestChoice. http://bit.ly/1eXggG0.
Beddow, J. M., D. Kriticos, P. G. Pardey, and R. W. Sutherst. 2010. Potential Global Crop Pest
Distributions Using Climex: HarvestChoice Applications. Washington, DC, and St. Paul,
MN, US: HarvestChoice. http://bit.ly/1hCwlGh.
Beddow, J., R. W. Sutherst, D. Kriticos, E. Duveiller, and Y. Chai. Oct. 2013a. Puccinia graminis (Stem Rust). Pest Geography. St. Paul, MN, US: HarvestChoice-InSTePP (International
Science & Technology Practice & Policy), University of Minnesota; and Canberra:
Commonwealth Scientific and Industrial Research Organisation.
Pardey, P. G., J. M. Beddow, D. J. Kriticos, T. M. Hurley, R. F. Park, E. Duveiller, R. W. Sutherst,
J. J. Burdon, and D. Hodson. 2013. Right-Sizing Stem-Rust Research. Science 340 (6129):
147148.
BENEFITS OF TRYPANOSOMOSIS CONTROL

IN THE HORN OF AFRICA
Cecchi, C., W. Wint, A. Shaw, A. Marletta, R. Mattioli, and T. P. Robinson. 2010. Geographic
Distribution and Environmental Characterization of Livestock Production Systems in
Eastern Africa. Agriculture, Ecosystems and Environment 135: 98110.
Shaw, A. P. M., G. Cecchi, G. R. W. Wint, R. C. Mattioli, and T. P. Robinson. 2014. Mapping
the Economic Benefits to Livestock Keepers from Intervening against Bovine
Trypanosomosis in Eastern Africa. Preventative Veterinary Medicine 113 (2): 197210.
http://bit.ly/1c52W7T.
Wint, G. R. W. 2001. Kilometre Resolution Tsetse Fly Distribution Maps for the Lake Victoria
Basin and West Africa. Report to the Joint Food and Agriculture Organization of the
United Nations (FAO)/International Atomic Energy Agency (IAEA) Programme. Vienna,
Austria: FAO\IAEA Joint Division.
ATLAS
Access to Trade
Market Access 66
Accessing Local Markets: Marketsheds 68
Accessing International Markets: Ports and Portsheds 70
Works Cited 72
65
ACCESS TO TRADE
Market Access
Zhe Guo and Cindy Cox

Most Africans do not have easy access to markets. To reach
a city of 50,000 people, a farmer in western Africa may only
have to travel 1 to 2 hours, whereas farmers in less densely
populated areas such as eastern Angola may need to travel
8 hours or more. The maps show travel time to major settlements with populations of 20,000 or more (Map 1), 50,000
or more (Map 2), 100,000 or more (Map 3), and 250,000 or
more (Map 4). Travel time is a proxy for accessibility and
shows how likely farming households are to be physically
integrated with or isolated from markets. Travel time is influenced not only by distance but also by infrastructure quality and road conditions. For example, because South Africa
has better infrastructure and more well-maintained roads
than the Democratic Republic of the Congo, it would take a
South African farmer less time to travel the same distance to
a market than a Congolese farmer. Another factor in determining market accessibility is the density of large cities in a
country. A country with many large cities, like Nigeria, has
highly accessible markets.

Improved market access for the poorest countries is widely
regarded as necessary to support agricultural and rural
development. In Africa the practice of trading agricultural
products is highly constrained by agricultural policies and
poor transportation networks. Challenging road conditions, long distances, and inadequate road infrastructure
add to travel times and transportation costs and therefore limit opportunities for farmers to sell their goods. Poor
market access can also negatively impact farm production,
because the accessibility of critical agricultural inputs such
as fertilizer, pesticides, and seed is also limited. Compared
to urban households and those with easy access to markets, rural farm households without market access typically rely on their own production for most of their calorie
intake. Inadequate market access, therefore, puts these
households at greater risk of food insecurity. The more
accessible markets are, the greater the populations ability
to remain economically self-sufficient and maintain food
66
security. A comparison of the maps, which express travel

time to different-sized cities (market centers), can help stakeholders better understand factors that determine farm performance. A simple cost-benefit analysis reveals whether it is
more profitable to travel longer distances to larger markets
or travel shorter distances to reach the nearest market.

Accessibility was determined using a cost-distance function
to measure the cost in hours to the nearest market center for each location, or 1 km2 grid cell. Market centers and
their size were determined using population estimates from
Global Rural Urban Mapping Project data for the year 2000
(CIESIN et al. 2011). Travel time was estimated based on the
combination of global spatial data layers, including road and
river networks, assessed in terms of their friction or kilometers per hour travel time. Travel time was adjusted based
on a number of input variables, including roadlocation, road
type, elevation, slope, country boundaries, bodies of water,
coastline, and land cover. Each input variable was converted
to a value representing the time it takes to travel 1 km. In
the case of road type, for example, paved roads were given
a value of 60 km per hour, while gravel roads were given a
value of 15 km per hour. Bodies of water, land cover, slope,
country boundaries, and elevation were also used to modify
the speed of travel. For example, steeper areas were assigned
slower speeds and time delays were factored into travel that
crossed borders. The results are not meant to be accurate
travel times but to estimate accessibility.

Market access:
http://harvestchoice.org/topics/market-access
Market access data for SSA:
http://harvestchoice.org/products/data/218
Global Rural-Urban Mapping Project population data:
http://bit.ly/KbKxJD
ATLAS
Market access based on population size of market centers

MAP 1
Population 20,000
MAP 2
Population 50,000
Travel time to
markets (hours)
0
1
2
4
6
8
12
Travel time to
markets (hours)
0
1
2
4
6
8
12
No data
No data
MAP 3
Population 100,000
MAP 4
Population 250,000
Travel time to
markets (hours)
0
1
2
4
6
8
12
Travel time to
markets (hours)
0
1
2
4
6
8
12
No data
No data
Data source: Map 1HarvestChoice 2011a; Map 2HarvestChoice 2011b;

Map 3HarvestChoice 2011c; Map 4HarvestChoice 2011d.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
67
ACCESS TO TRADE
Accessing Local Markets: Marketsheds
Zhe Guo and Cindy Cox

Across Africa buying and selling connects people. For
a small-scale farmer, this trade takes place primarily within
a limited geographic area based on access to market centers of a given size. The maps illustrate these areas using
different colors to represent marketshedsgeographical areas and associated populations that are part of real
or potential trade networks with a given market. From any
location within a marketshed, it takes less time to travel
to the corresponding market compared to any neighboring markets. In theory, farmers within a marketshed prefer
to trade their commodities at the corresponding market,
which minimizes travel cost (p. 66). The maps show that
the density of marketsheds in Nigeria is high compared to
that of other countries, because the country has many large
cities. The high concentration of marketsheds also shows
that it takes less time to travel to markets in Nigeria compared to neighboring countries. This suggests a denser and
perhaps higher-quality infrastructure. The progression of
Maps 14 shows that as the size of market centers, based
on population, increases, there are fewer markets across the
continent. Farmers thus have to travel farther, often across
country boundaries, to reach larger market centers which
may represent more lucrative trade opportunities.

When analyzing factors that influence current and future
farm performance, development planners and researchers
need to know which markets are closest to agricultural producers. Farmers customarily select markets close to them so
they can get to the market in the least amount of time to
trade their goods; buy critical agricultural inputs, such as fertilizer, seed, and pesticides; or tap into a range of public and
private services (extension, credit, and veterinary services
being prime examples). A relatively large marketshed could
mean that the population density for that shed is so low that
few markets exist, and therefore that farmers have limited
opportunities to sell their products (such as in Namibia). Or
it might mean that the market within the shed serves a large
population most likely due to adequate investments in road
infrastructure. The maps show that the marketsheds are
68
not restricted by country borders, which means that a farmers preferred market of a given size may be in a neighboring
country. In that case, restrictions posed by border crossings
and trade laws need to be considered when determining the
optimal market for a farmer. Because each map is based on
market centers of different sizes, they can be used to determine the best markets for selling a farmers goods. Farmers
with an abundance of high-value goods will often prefer to
sell or trade at larger commercial markets where demand
and prices are higher than at smaller local markets.

Marketsheds are based on the cost of travel to a market
center of a given size. The number of marketsheds in a
country indicates the number of market centers of that
size within the country (for example, Map 1 is based on
a market-center population of 50,000 or greater). The population cutoffs used in the maps are based on population estimates from Global Rural-Urban Mapping Project (GRUMP)
data for the year 2000 (CIESIN et al. 2011).Proximity to a
marketwas determined by measuring the lowest accumulated cost, or travel time, to each market location. Every market is surrounded by a marketshed. All points within the
marketshed area offer the shortest travel time to the corresponding market center. Points along the boundary between
two sheds have equal travel time to both of the centers.
Travel time is estimated based on a combination of spatial data layers and variables that affect the time required to
travel to the cities or market centers. These variables include
elevation, slope, land cover, roads, road types, rivers, borders,
and major bodies of water (Guo 2010).

Marketsheds for Africa south of the Sahara (SSA):
http://harvestchoice.org/labs/market-sheds
Market access:
http://harvestchoice.org/topics/market-access
Marketshed data for SSA: http://bit.ly/1oFyB1B
ATLAS
Marketsheds based on population size of market centers

MAP 1
Population 50,000
MAP 2
Population 100,000
MAP 3
Population 250,000
MAP 4
Population 500,000
Data source (all maps): HarvestChoice 2012.

Note: Population data used are for the year 2000 (CIESIN 2011). The different colors represent
marketsheds. A marketshed is the total area surrounding a market center of a given size. From
any point within the marketshed, it is quicker to travel to that market center than to any
neighboring marketsheds main market.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
69
ACCESS TO TRADE
Accessing International Markets: Ports and Portsheds
Zhe Guo

More than 300 million Africans, about 30 percent of the
total population, live more than one day away from the nearest port. Even when ports lie within a few hundred miles,
typically sparse road networks, poor maintenance, and limited transportation infrastructure translate into high access
costs. The larger map illustrates cost-of-travel accessibility to
63 major African ports, based on port type, size, and capacity in terms of the estimated total number of hours, both off
and on the road network, required to travel from any location in Africa to the nearest port. The populations, traders,
and haulage operations of countries such as South Africa
and Egypt that maintain more and better ports as well as
better transportation infrastructure have significantly better port access than those in landlocked countries such as
Chad and South Sudan or large countries such as Democratic
Republic of the Congo where infrastructure is limited. The
travel time analysis underpinning the map is further summarized in Map 2, which shows portsheds. A portshed is a
ports catchment area. Each portshed includes all the locations that are closer to a given port in terms of travel time
than to any other port. Ports with large catchment areas,
such as Mombasa in Kenya, have few competing ports and
are connected to more extensive road networks. Ideally
each port should be endowed with transportation corridors,
infrastructure, and port facilities that maximize the trading
opportunities within its specific portshed.

Seaports play a significant role in enabling both export opportunities for agricultural products and import potential for
new technologies and production inputs. Indeed, more than
90 percent of the international trade in African countries is
conducted using maritime transport. Most African countries
import vital agricultural inputs such as fertilizers, seeds, pesticides, and herbicides. Crops (especially cash crops) and livestock products (including skins and hides and, in the Horn of
Africa, live animals) are primary agricultural exports. Map 1,
showing travel time, provides a picture of how isolated many
Africans are from such import and export hubs that could
connect them with world markets and expand their earning
potential. This information is valuable to policymakers and
investors, both public and private. It allows them to identify
70
TABLE 1 Distribution of major African ports by region

and size
REGION
PORT SIZE
TOTAL
Large
Medium
Small
Eastern Africa
14
Middle Africa
Northern Africa
15
22
Southern Africa
Western Africa
16
Total
39
22
66
Data source: World Port Source 2012 and FAO 2012.

Note: The classification of harbor size is based on several applicable factors,
including area, facilities, and wharf space. It is not based on area alone, nor any
other single factor (National Geospatial-Intelligence Agency 2012).
intervention priorities that will, assuming sufficient competition in the transportation sector, reduce transaction costs and
increase the capacity and efficiency of transportation systems.
This ultimately improves production incentives for farmers
and raises farm-level productivity and profitability by lowering input costs and increasing output prices.

Travel time was estimated using a combination of spatial
data layers and variables that influence accessibility, including elevation, slope, land cover, the road network, road types,
rivers, borders, and major bodies of water. Esris ArcGIS
Spatial Analyst was used to develop spatial indicators of
travel time to 63 major ports in Africa, which were selected
based on port size and regional distribution (Table 1). The
continent was then divided into portsheds, each defining
the area associated with the closest corresponding port.
The closest port was determined by estimating the lowest
accumulated travel time (or cost) from a geographic location to the port. Using this approach, port A is the closest
port for any geographic location within portshed A.

Portshed data: http://bit.ly/1eRgKkI
World Port Source: www.worldportsource.com
ATLAS
MAP 1
Travel time to major African ports
Hours
0
4
12
24
48
Major African port
MAP 2 Portsheds for 63 major

African ports
Major African port
Data source: Map 1HarvestChoice 2012 and World Port Source 2012;
Map 2HarvestChoice 2012.
Note: Map 2The different colors represent portsheds based on access to a major
port. A portshed is the total area surrounding a major port for which the given port
is closer in terms of travel time than any other port.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
71
ACCESS TO TRADE
Works Cited
MARKET ACCESS
ACCESSING LOCAL MARKETS: MARKETSHEDS
CIESIN (Center for International Earth Science Information Network)/Columbia University,

International Food Policy Research Institute, the World Bank, and Centro International
de Agricultura Tropical. 2011. Global Rural-Urban Mapping Project, Version 1
(GRUMPv1): Settlement Points. http://bit.ly/1fgDVBC.
HarvestChoice. 2011a. Average Travel Time to Nearest Town Over 20K (hours) (2000).
International Food Policy Research Institute and University of Minnesota. Accessed
January 27, 2014. http://harvestchoice.org/node/5210.
. 2011b. Average Travel Time to Nearest Town Over 50K (hours) (2000).
. 2011c. Average Travel Time to Nearest Town Over 100K (hours) (2000).
. 2011d. Average Travel Time to Nearest Town over 250K (hours) (2000).
HarvestChoice. 2012. Market Sheds. http://harvestchoice.org/labs/market-sheds.

CIESIN (Center for International Earth Science Information Network)/Columbia University,
International Food Policy Research Institute, the World Bank, and Centro International
de Agricultura Tropical. 2011. Global Rural-Urban Mapping Project, Version 1
(GRUMPv1): Settlement Points. http://bit.ly/1fgDVBC.
72
ACCESSING INTERNATIONAL MARKETS:

PORTS AND PORTSHEDS
FAO (Food and Agriculture Organization of the United Nations). 2012. FAOSTAT database.
http://bit.ly/1dUsYWv.
HarvestChoice. 2012. Infrastructure and Transportation. http://bit.ly/1dhJ7s9.
National Geospatial-Intelligence Agency. 2012. World Port Index, 22nd ed. Pub. 150.
Springfield, VA, US.
World Port Source. 2012. www.worldportsource.com/index.php.
ATLAS
Human Welfare
Severity of Hunger 74
Poverty 76
Early Childhood Nutrition and Health 78
Works Cited 80
73
HUMAN WELFARE
Severity of Hunger
Klaus von Grebmer, Tolulope Olofinbiyi, Doris Wiesmann, Heidi Fritschel, Sandra Yin, and Yisehac Yohannes
Map 1 shows the severity of hunger in Africa by categoriesranging from low to extremely alarming. These categories are associated with Global Hunger Index (GHI) scores.
Higher scores indicate greater hunger; the lower the score,
the better a countrys situation. Of the 19 countries worldwide with alarming or extremely alarming levels of hunger,
most (15) are in Africa south of the Sahara. Map 2 shows
country progress in reducing GHI scores since 1990that is,
the percentage change in the 2013 GHI compared with the
1990 GHI. An increase in the GHI indicates a countrys hunger situation is deteriorating. A decrease in the GHI indicates
an improvement.
Overall, from the 1990 GHI to the 2013 GHI, six countries in Africa were able to reduce their scores by 50 percent
or more. Twenty countries made modest progress, reducing
their GHI scores by 25.0 to 49.9 percent, and 17 countries
decreased their GHI scores by 0.0 to 24.9 percent. Hunger
grew worse in Burundi, Comoros, and Swaziland (Map 2).
Increased hunger in Burundi and Comoros can be attributed
to prolonged conflict and political instability. For Burundi,
the share of undernourished people in the population rose
from 49 to 73 percent between the 1990 GHI and 2013 GHI.
In Swaziland (Figure 1), the HIV and AIDS epidemic, along
with high unemployment and adverse macroeconomic conditions, likely undermined food security. Ghana, the top
performer in Africa in terms of improved GHI scores since
1990 (Figure 1), is the only country in Africa to appear on
the top 10 list worldwide. Significant drops in the share of
undernourished population and in the prevalence of underweight in children under five (p. 78) contributed to Ghanas
2013 GHI of 8.2, down from the 1990 GHI of 25.5 (Figure 1).

The GHI is designed to comprehensively measure and track
hunger globally, by country, and by region. It highlights successes and failures in reducing hunger and provides insights
into its drivers. By highlighting regional and country differences, the GHI aims to trigger actions to reduce hunger. The
GHI is a multidimensional index of hunger that combines
three equally weighted indicators (undernourishment, child
underweight, and child mortality) in one number. This multidimensional approach takes into account the nutrition situation not only of the population as a whole, but also of a
physiologically vulnerable groupinfants and young childrenfor whom a lack of nutrients (p. 78) creates a high
74
FIGURE 1
Trends in GHI scores for two countries
30
GHI 1990
GHI 1995
25
GHI 2000
GHI 2005
20
GHI 2013
15
10
0
Ghana
Swaziland
Source: von Grebmer et al. 2013.
risk of illness, poor physical and cognitive development,

and death.

The 2013 GHI was calculated for 120 countries globally for
which data were available and where measuring hunger is
considered most relevant. The GHI is only as current as the
data for the three component indicators: undernourishment,
child underweight, and child mortality. Source data for the
2013 GHI are from 2008 to 2012 (von Grebmer et al. 2013).
Therefore, the GHI is a snapshot of the recent past, not the
present. More up-to-date and extensive country data on
hunger are urgently needed. The Democratic Republic of the
Congo, for example, had the worst score in past GHI reports.
But due to political instability and ongoing conflict, reliable
data are no longer available to calculate its GHI.

2013 Global Hunger Index: http://bit.ly/KaKqhr
ATLAS
MAP 12013 Global Hunger Index scores
MAP 2 Percentage change in 2013 GHI compared with

1990 GHI
Tunisia
Tunisia
Morocco
Morocco
The
Burkina
Faso
Gambia
Guinea
Nigeria
GuineaCte
Bissau
d'Ivoire Ghana
Central African
Sierra
Republic
Cameroon
Leone
Togo Benin
Liberia
Democratic
Sao Tome and
Republic
Principe
of the Congo
Gabon
Equatorial
Rwanda
Guinea
Burundi
Republic
Severity: score
of Congo
Low: 4.9
Moderate: 5.09.9
Serious: 10.019.9
Angola
Somalia
Ethiopia
Ug
da
an
Kenya
Tanzania
Malawi
Seychelles
Comoros
Zambia
Zimbabwe M
a
oz
u
bi q
Namibia Botswana
Alarming: 20.029.9
Extremely alarming: 30.0
Djibouti
Swaziland
South Africa
Lesotho
No data
Data source (all maps): von Grebmer et al. 2013.

Note: The 2013 Global Hunger Index score could only be calculated for former Sudan,
because separate undernourishment estimates for 20102012 were not available for
(north) Sudan or South Sudan, which became independent in 2011.
Mali
Eritrea
Niger
Senegal
Chad
Sudan
(former)
The
Burkina
Faso
Gambia
Guinea
Nigeria
GuineaCte
Bissau
d'Ivoire Ghana
Central African
Sierra
Republic
Cameroon
Leone
Togo Benin
Liberia
Democratic
Sao Tome and
Republic
Principe
of the Congo
Gabon
Equatorial
Rwanda
Guinea
Burundi
Republic
of Congo
Decrease by 50.0% or more

Decrease by 25.049.9%
Decrease by 0.024.9%
Angola
Somalia
Ethiopia
Ug
da
an
Kenya
Tanzania
Malawi
Seychelles
Comoros
Zambia
Zimbabwe M
a
oz
u
bi q
Namibia Botswana
Increase
1990 & 2013 GHI <5
Djibouti
c ar
Sudan
(former)
agas
Chad
Egypt
Algeria
Mauritania
Eritrea
Niger
c ar
Mali
agas
Mauritania
Senegal
Libya
Western
Sahara
Egypt
Algeria
M ad
Western
Sahara
M ad
Libya
Swaziland
South Africa
Lesotho
No data
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
75
HUMAN WELFARE
Poverty
Carlo Azzarri

Poverty prevalence (Maps 1 and 3) is crucial information
for policymakers and international donors who are setting
priorities for intervention and investment. Poverty density
complements prevalence by showing the number of poor
people per square kilometer (Maps 2 and 4). These maps
together answer two important questions: Where is poverty
a serious problem? Where might investments have the greatest impact on the highest number of people? Combining
insights on both prevalence and density allows policymakers
to more effectively target interventions to reach the greatest number of the poorest people. Once target populations
are identified, information on the dominant types of existing livelihoods and agriculture-related opportunities can be
helpful in formulating appropriate interventions.
Poverty headcount ratio by agroecological zone
Subtropic
Almost half of the population of Africa south of the Sahara

(SSA) lives in extreme poverty, on less than $1.25 per capita per day.1 Map 1 shows the distribution of the poor and
highlights areas where over 80 percent of the population
is extremely poor (for example, parts of Liberia, Nigeria,
Tanzania, and Zambia). Map 2 shows the density of extremely
poor across the continent, highlighting regions that are home
to more than 100 extremely poor people per square kilometer. Moderate poverty is defined as living on a daily per capita expenditure between $1.25 and $2.00. Map 3 shows the
distribution of poor using the $2.00 per day threshold, thus
including both the moderately and extremely poor. This map
shows a more even distribution of poor across Africa and
consistently higher shares of the total population. Map 4 reinforces that the most densely populated poor areas are concentrated along the coast of western Africa, in much of
Nigeria, in Malawi, in Ethiopia, and in the countries bordering
or near Lake Victoria. Figure 1 shows that extreme poverty
is also highly correlated with certain agroecological zones
(p. 34). For example, poverty levels are highest in the warm
semiarid and subhumid tropical areas immediately south
of the Sahara and in the tropical warm humid forests of the
Democratic Republic of the Congo. And, overall, poverty levels are lower in the subtropical zones of southern Africa (for
example, Namibia and South Africa).
FIGURE 1
Warm arid
Cool subhumid
Cool semiarid
Warm semiarid
Warm subhumid
Cool arid
Warm humid
Cool humid
Tropic
Warm semiarid
Warm subhumid
Warm humid
Cool subhumid
Cool humid
Cool semiarid
Warm arid
Cool arid
0
20
40
Poverty headcount ratio (%)
60
Data source: HarvestChoice 2012 and HarvestChoice 2011.

Note: Poverty headcount ratio=the percentage of a population living in households where consumption or income per person is below the poverty line.

Subnational poverty rates were extracted from 24 nationally
representative household surveys conducted in various years.
For countries without survey data, national average poverty
prevalence extracted from PovcalNet (World Bank 2012) for
the closest year to 2005 was uniformly applied to the entire
country. As such, subnational poverty rate distributions reflect
the relative ranking in the actual survey year, although all values are expressed in terms of 2005 average purchasing power
parity exchange rates. Poverty ratios are therefore comparable across countries. Not all current data points refer to 2005,
with a maximum variance of plus or minus two years for a
limited number of countries (HarvestChoice 2012).

Poverty analysis at the World Bank:
www.worldbank.org/en/topic/poverty
Poverty Comparisons over Time and Across Countries in
Africa. Sahn and Stifel 2000.
Where Will the Worlds Poor Live?: An Update on Global
Poverty and the New Bottom Billion. Sumner 2012.
1 The $1.25 and $2.00 poverty lines are the level of total household per capita consumption expenditure (a synthetic indicator of household welfare) expressed
in terms of 2005 average purchasing power parity exchange rates.
76
ATLAS
MAP 1 Share of population living at $1.25/day

(extremely poor)
MAP 2
Poverty density at $1.25/day (extremely poor)
Prevalence (percent)
1
10
20
30
40
50
60
70
80
90
100
No data
Outside focus area
MAP 3 Share of population living at $2.00/day (includes

moderately and extremely poor)
Number of poor/km
0
5
10
25
50
100
500
No data
Outside focus area
MAP 4 Poverty density at $2.00/day

(includes moderately and extremely poor)
Prevalence (percent)
1
10
20
30
40
50
60
70
80
90
100
No data
Outside focus area
Data source (all maps): HarvestChoice 2012.

Note: All values are expressed in terms of average 2005
purchasing power parity rates.
Number of poor/km
0
5
10
25
50
100
500
No data
Outside focus area
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
77
HUMAN WELFARE
Early Childhood Nutrition and Health
Carlo Azzarri

High levels of stunting, or lower than average height in
children younger than five, are more widespread in Africa
south of the Sahara (SSA) than high levels of wasting (lower-than-average weight for height) or underweight (low
weight for age) in children under age five (Maps 1, 2, and
3). This reflects a longstanding nutritional problem that
has proven difficult to eradicate in this region. Even with
improved living conditions in SSA, the prevalence of stunting
has not yet been sufficiently reduced. Stunting and underweight are manifestations of undernutritionfood energy
deprivation that occurs when food intake is below standard
nutritional requirements for a prolonged period and/or levels of food absorption are low. Wasting usually reflects an
acute weight loss due to a recent period of hunger or disease
and is often associated with shorter term limitations to food
supplies. The maps show that high rates of undernutrition
do not always correspond to high rates of diarrhea (Map 4),
which contribute to undernutrition by interfering with the
absorption of food consumed. This suggests that poor infrastructure and lack of access to clean water (the main causes
of diarrhea) are just two of many reasons behind the severe
undernutrition in SSA. The red areas of the maps reflect
undernutrition levels classified as very high40 percent or
above for stunting; 15 percent or above for wasting; 30 percent and above for underweight (WHO 2006); and 20 percent or above for diarrheaand highlight the key areas for
concern across the continent.
impairing their cognitive development and affecting their

performance once they are adults. Better nutrition translates into a stronger and healthier population with greater
opportunities to break the cycle of poverty and achieve better quality of life. Improving childrens nutritional status is
therefore fundamental to realizing a countrys development
potential, especially in nations in SSA where nearly half of
the population is less than 15 years old.

Measurements are usually taken from children from birth
up to 60 months, as this captures the impact of possible
deficiencies incurred during gestation, and it is when children are most vulnerable as they rapidly grow and develop.
After the 1,000-day window of opportunity (from the start
of a womans pregnancy until her childs second birthday),
any impaired height development or cognitive function is
largely irreversible. To obtain anthropometric measures, we
used the childrens weight, height, and age information collected in the Demographic and Health Survey (DHS) Phase 5
(20032008) and Phase 6 (20082013). The DHS surveys are
regularly conducted in many developing countries in different years, and these maps show the values for countries with
survey years ranging from 2003 to 2011 (Measure DHS 2013).

Measure DHS online: www.statcompiler.com/

The information on these maps is crucial to policymakers and national and international donors who seek to
direct resources to the most food-insecure regions of the
world. Child nutrition is often used as an indicator of an
areas nutrition security. According to the World Health
Organization (WHO), child undernutrition is directly or
indirectly responsible for one-third of the deaths among
children under age five, and it is also related to other illnesses common in children, such as diarrhea and measles.
Undernutrition carries long-term consequences for children,
78
WHO Child Growth Standards Publications:

www.who.int/childgrowth/publications/en/
Explaining Child Malnutrition in Developing Countries:
A Cross-Country Analysis. Smith and Haddad 2000.
Poverty and Undernutrition: Theory, Measurements, and
Policy. Svedberg 2000.
Worldwide Timing of Growth Faltering: Revisiting
Implications for Interventions. Victora et al. 2010.
ATLAS
Nutrition and health among children under age five

MAP 1Stunting
Prevalence: percent
MAP 2Wasting
Prevalence: percent
Low: < 20
Low: < 5
Medium: 2029
Medium: 59
High: 3039
High: 1014
Very high: 40
Very high: 15
No data
No data
Outside focus area
Outside focus area
MAP 3Underweight
Prevalence: percent
MAP 4Diarrhea
Prevalence: percent
Low: < 10
Low: < 11
Medium: 1019
Medium: 1115
High: 2029
High: 1520
Very high: 30
Very high: 20 <
No data
No data
Outside focus area
Outside focus area
Data source (all maps): Measure DHS 2013 and WHO 2006.
Note: The maps are based on DHS surveys conducted over the period 2003 to 2011.
The maps show prevalence classes and corresponding undernutrition levels (as a share
of total children under age five) as designated by the World Health Organization.
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
79
HUMAN WELFARE
Works Cited
SEVERITY OF HUNGER
EARLY CHILDHOOD NUTRITION AND HEALTH
von Grebmer, K., D. Headey, C. Bn, L. Haddad, T. Olofinbiyi, D. Wiesmann, H. Fritschel,

S. Yin, Y. Yohannes, C. Foley, C. von Oppeln, and B. Iseli. 2013. 2013 Global Hunger Index:
The Challenge of Hunger: Building Resilience to Achieve Food and Nutrition Security.
Bonn, Germany: Welthungerhilfe; Washington, DC: International Food Policy Research
Institute; Dublin: Concern Worldwide.
Measure DHS (Demographic and Health Surveys). 2013. STATcompiler. Accessed November
11, 2013. http://www.statcompiler.com/.
Smith, L. C., and L. Haddad. 2000. Explaining Child Malnutrition in Developing Countries:
A Cross-Country Analysis. Washington, DC: International Food Policy Research Institute.
Svedberg, P. 2000. Poverty and Undernutrition: Theory, Measurements, and Policy. Oxford,
UK: Oxford University Press.
Victora, C. G., M. de Onis, P. C. Hallal, M. Blssner, and R. Shrimpton. 2010. Worldwide
Timing of Growth Faltering: Revisiting Implications for Interventions. Pediatrics 125 (3):
473480.
WHO Multicentre Growth Reference Study Group. 2006. WHO Child Growth Standards:
Length/Height-for-Age, Weight-for-Age, Weight-for-Length, Weight-for-Height and Body
Mass Index-for-Age: Methods and Development. Geneva: World Health Organization.
http://bit.ly/1i7JPfX.
POVERTY
HarvestChoice. 2011. AEZ (16-class). http://harvestchoice.org/data/aez-16-class.
. 2012. Sub-national and Extreme Poverty Prevalence. International Food Policy
Research Institute and University of Minnesota. Accessed January 27, 2014.
http://harvestchoice.org/node/4751.
Sahn, D. E., and D. Stifel. 2000. Poverty Comparisons Over Time and Across Countries in
Africa. World Development 28 (12): 21232155.
Sumner, A. 2012. Where Will the Worlds Poor Live? An Update on Global Poverty and the
New Bottom Billion. World Development 40 (5): 865877.
World Bank. 2012. PovcalNet: An Online Poverty Analysis Tool. http://iresearch.
worldbank.org/PovcalNet/index.htm.
80
About the Authors

Christopher Auricht (chris@auricht.com) is managing director of Auricht Projects, a consultancy in Adelaide, South Australia. His background is in applied science in natural resource
management. He has extensive international experience in the development and implementation of programs and frameworks that help provide quality data and information for planning for natural resources and sustainable management and policymaking.
Carlo Azzarri (c.azzarri@cgiar.org) is a research fellow on the HarvestChoice research team
at the International Food Policy Research Institute (IFPRI), Washington, DC. His research
involves micro- and macroeconomic analysis, using quantitative (statistical and econometric)
and qualitative methods, of the interrelationships among poverty, nutrition, food security,
agriculture, and migration. Before joining IFPRI, Azzarri worked in the research group on poverty and inequality at the World Bank and was a member of the Rural Income Generating
Activities team at the Food and Agriculture Organization of the United Nations (FAO).
Jason Beddow (beddow@umn.edu) is an assistant professor of international agriculture at
the University of Minnesota (UMN), MinneapolisSt. Paul. His work focuses on how research
investment patterns and aspects of the natural environment (including pests, diseases,
climate, and the weather) affect agricultural production and productivity, the spatial dynamics of production, and efforts to bridge the divide between biology and economics.
Nienke Beintema (n.beintema@cgiar.org) is program head of the Agricultural Science and
Technology Indicators (ASTI) initiative at IFPRI, Washington, DC. She has close to 20 years of
experience doing agricultural research and development in low- and middle-income countries. Her work has involved collecting and analyzing financial and human resource data and
examining institutional developments.
Chandrashekhar Biradar (c.biradar@cgiar.org) is an agroecosystem ecologist who heads
Geoinformatics and is the principal scientist at the International Center for Agricultural
Research in the Dry Areas (ICARDA). His current research interests include applying
geoinformatics to agroecosystem research and the remote sensing of global food and
environmental security in dry areas of the world.
Jean-Marc Boffa (j.m.boffa@cgiar.org) is a farming systems project manager and consultant
at the World Agroforestry Centre (ICRAF), Nairobi. An experienced systems agronomist with
wide experience in western, central, and eastern African farming systems, he has worked for
ICRAF in various roles and is a recognized authority on parkland agroforestry systems.
Giuliano Cecchi(giuliano.cecchi@fao.org) is a project officer for FAO, Rome. He was trained
as an environmental engineer. He specializes in geospatialanalysis, data management, and
remote sensing. Over the last eight years with FAO, he has focusedon the epidemiology and
biogeography of tsetse-transmitted Africantrypanosomoses.
Yuan Chai (chaix026@umn.edu) is a PhD student in the department of applied economics
at UMN, MinneapolisSt. Paul. Chai holds an MS in plant pathology and has extensive experience working with rust diseases. His research focuses on the bioeconomics of agricultural
pests and diseases.
Giuseppina Cinardi(giuseppina.cinardi@fao.org) works as a livestock information analyst
for the Livestock Information, Sector Analysis and Policy Branch of FAO, Rome. She organizes,
updates, and maintains a comprehensive database of livestock information and statistics.
About the Authors81
Lieven Claessens (l.claessens@cgiar.org) is a principal scientist specializing in natural resources

(soil and water) at the International Crops Research Institute for the Semi-Arid Tropics
(ICRISAT), Nairobi, and an assistant professor in soil geography and landscape at Wageningen
University, the Netherlands. His current work focuses on spatial analysis, integrated assessment, and modeling of agricultural systemswith an emphasis on smallholder systems in
Africa south of the Sahara and on adaptation and mitigation strategies in the context of climate change.
Giulia Conchedda (giulia.conchedda@fao.org) is a livestock and livelihoods information
analyst in the Livestock Information, Sector Analysis and Policy Branch of FAO, Rome. She
has many years of experience as a spatial data analyst and is particularly interested in socioecological approaches to studying livestock production systems and associated livelihoods.
Cindy Cox (c.cox@cgiar.org) is a technical writer on the HarvestChoice team at IFPRI,
Washington, DC. Cox holds an MS and PhD in plant pathology. Before joining IFPRI, she
worked as a Peace Corps volunteer in the Central African Republic and as an agricultural scientist at The Land Institute.
John Dixon (John.Dixon@aciar.gov.au) is a principal advisor in the research division of the
Australian Centre for International Agricultural Research, Canberra, Australia. He has over
40 years experience working for the CGIAR system and FAO on agricultural research and
development, including cropping systems, economics, and natural resource management in
South, Southeast, and East Asia; Africa; Latin America; and the Middle East. He has served
as director of Impacts, Targeting and Assessment at the International Maize and Wheat
Improvement Center (CIMMYT), leading activities on impact assessment, value chains,
impact knowledge sharing, systems agronomy, and conservation agriculture. He also served
in various capacities with FAO in its global, regional, and country programs.
Petra Dll (p.doell@em.uni-frankfurt.de) is a professor of hydrology at the Goethe
University Frankfurt in Frankfurt am Main, Germany. Her research focuses on the global
freshwater system and on interdisciplinary research in natural resources management. Dll
is co-developer of WaterGAP, a model of global water resources and their use that has been
used to assess the anthropogenic impact on the global freshwater system, including the
impact of climate change, reservoirs, and water withdrawals from groundwater and surface water.
Kathleen Flaherty (k.flaherty@cgiar.org) is a senior research analyst with the ASTI initiative at IFPRI, Washington, DC. Her work focuses on collecting and analyzing human
and financial resource data related to agricultural research and development in lowand middle-income countries.
Karen Frenken (karen.frenken@fao.org) is the senior officer for Water Resource Management at the Land and Water Division of FAO, Rome. She worked for almost 20 years as an
agricultural engineer in different countries in South Asia, the Near East, and Africa south of
the Sahara, mainly on irrigation and water management for agricultural purposes. A particular focus of her work was fragile ecosystems with special attention to women and irrigated
agriculture. In 2003, she joined FAO headquarters, where she is responsible for managing the
AQUASTAT Programme, FAOs global water information system.
Heidi Fritschel (h.fritschel@cgiar.org) is an editor in the Communications and Knowledge
Management Division of IFPRI, Washington, DC.
82
Atlas of African Agriculture Research & Development
Dennis Garrity (d.garrity@cgiar.org) is a senior fellow at the World Agroforestry Centre

(ICRAF), Nairobi. He is a systems agronomist and research leader whose career has focused
on the development of small-scale farming systems in the tropics. He is currently serving as
drylands ambassador for the United Nations Convention to Combat Desertification, raising
awareness about the importance of combating desertification and land degradation and mitigating the effect of drought. He is also involved in a global effort to reconsider the future of
agriculture in the 21st century by examining unconventional ways of creating more productive and environmentally sound farming.
Marius Gilbert(Marius.Gilbert@ulb.ac.be) is a fellow of theFonds National de la Recherche
Scientifique, which is based in the Department ofBiological Control and Spatial Ecology at
the Universit Libre de Bruxelles in Brussels, Belgium. He has many years of experience in spatial ecology and epidemiology, and his recent research has focused onmodelingof livestock
distributions, production systems, and both the distributions and systems associations with
disease risk.
Zhe Guo (z.guo@cgiar.org) is a geographic information system coordinator in the Environ
ment and Production Technology Division of IFPRI, Washington, DC, and is a member of the
HarvestChoice team. His research interests include using spatial data and remotely sensed
data for spatial modeling, spatial statistics, and evaluating land cover and land use changes.
Jawoo Koo (j.koo@cgiar.org) is a research fellow in the Environment and Production Technology
Division of IFPRI, Washington, DC, and is on the HarvestChoice team. He specializes in crop
modeling. His current work explores strategic questions about agricultural technologies in
Africa, including which areas to target with the technologies and their potential impacts.
Raffaele Mattioli(raffaele.mattioli@fao.org) works as a senior officer in the Animal Health
Division of FAO, Rome. He is aveterinarian with more than 25 years of experience in tropical
animalhealth and production, with a focus on tsetse and trypanosomosis interventionsand
other programs that control vectors and vector-borne diseases. He now leadsthe disease
ecology group.
Tolulope Olofinbiyi (t.olofinbiyi@cgiar.org) is a program manager in the Director Generals
Office of IFPRI, Washington, DC. She is also a PhD candidate studying development economics and political economy at the Fletcher School at Tufts University. She holds a master of
arts in law and diplomacy in international affairs (development economics) and a master of
agribusiness and has extensive experience working in the agribusiness sector in Nigeria. Her
research focuses on the political economy of public finance in the context of fiscal federalism
and decentralization.
Felix T. Portmann (felix.portmann@senckenberg.de) is a postdoctoral research associate at
the Biodiversity and Climate Research Centre linked to the Senckenberg Research Institute
and Natural History Museum and Goethe University Frankfurt in Frankfurt am Main,
Germany. Before obtaining his PhD from Goethe University, he was a scientist at the German
Federal Institute of Hydrology. His research interests include environmental modeling of the
hydrological cycle, coupled climate and ocean modeling, statistical analysis related to past
and present climates, integrated water resources management, and remote sensing.
Navin Ramankutty (navin.ramankutty@mcgill.ca) is an associate professor of geography
at McGill University, Montreal. His research program addresses issues at the intersection of
global food security, land use change, and global environmental change.
About the Authors83
Timothy Robinson(t.robinson@cgiar.org) works as a senior spatial analyst at the

International Livestock Research Institute (ILRI),Nairobi. His interests include spatial analytical techniques and theirapplication to understanding current and future distributions
of livestockand farming systems. He is particularly interested in exploring the social,environmental, and epidemiological risks and opportunities associated withan evolving livestock sector.
Kate Sebastian (ksebconsult@gmail.com) is a consultant with the Bill & Melinda Gates
Foundation, Seattle, and the project manager of the eAtlas initiative. She has worked in the
field of geographic information systems and agriculture research for a number of organizations including IFPRI, the US Agency for International Development, the World Bank, the
HarvestChoice team, and the CGIAR Consortium. Her focus is on mapping and spatial analyses of data related to agricultural land use, poverty, and food security.
Alexandra Shaw(alex@apconsultants.co.uk) is an independent consultant, currently associated with theDivision of Pathway Medicine and Centre for Infectious Diseases at the
University of Edinburgh, UK. She is an economist with many years of experiencein the health
economics of the livestock sector, particularly in the study oftsetse-transmitted trypanosomosis and, more recently, of other neglectedzoonotic diseases.
Stefan Siebert (s.siebert@uni-bonn.de) is a senior scientist at the Institute of Crop Science
and Resource Conservation of the University of Bonn, Germany. His research focuses on the
analysis of large-scale datasets and the development of large-scale model components to
assess the interaction among resource use, crop management, and crop productivity. This
model development includes the generation and improvement of datasets required as model
inputs. Sieberts research interests also include general crop modeling (phenology, drought
and heat stress, and crop yields), modeling of virtual water flows, and the alteration of ecosystems by anthropogenic impacts.
Gert-Jan Stads (g.stads@cgiar.org) is senior program manager of the ASTI initiative at IFPRI,
Washington, DC. His focus is on conducting ongoing analysis of agricultural research and
development datasets, communicating the results of this analysis to promote advocacy and
support policymaking, and building in-country capacity for data collection and analysis.
Philip Thornton (p.thornton@cgiar.org) leads the Integration for Decision Making theme
for the CGIAR Research Program on Climate Change, Agriculture and Food Security, which is
hosted by ILRI in Nairobi. Thornton is based in Edinburgh, UK. He works mostly in integrated
modeling at different scales, looking at the impacts of climate change on smallholder farming systems in the tropics and subtropics.
Antonio Trabucco (antonio.trabucco@cmcc.it) is a researcher at the Euro-Mediterranean
Center on Climate Change, Lecce, Italy. He is a landscape ecologist with a background in
geography and a scientific interest in the interactions among climate, society, agriculture,
and natural ecosystems. Trabucco is currently investigating the impact of climate and climate
change on agricultural production and ecosystem services.
Justin Van Wart (s-jvanwar1@unl.edu) is a postdoctoral research associate in the
Department of Agronomy and Horticulture of the University of Nebraska, Lincoln. His work
focuses on understanding spatial and temporal patterns of agricultural management and
crop yields, estimating crop yield potential on the regional and national scales, and deriving globalclimate data for use in agroclimatology researchas well as for fieldcrop simulation models.
84
Klaus von Grebmer (k.vongrebmer@cgiar.org) is a research fellow and strategic advisor in

the Director Generals Office of IFPRI, Washington, DC. He served as director of IFPRIs
Communications Division for over 12 years. His work focuses on communicating complex
issues to selected stakeholder groups and on communication strategies, science communications, and issues management.
Doris Wiesmann (d.wiesmann@gmx.com) is an independent food security and nutrition
expert based in Halendorf, Germany. She is a nutritionist with extensive experience in measuring food security, designing and reviewing international indexes, and developing proxy
indicators of household food security and micronutrient adequacy. Wiesmann designed the
Global Hunger Index. More recently, her work has focused on dietary assessment methods
and strategies to improve micronutrient adequacy in developing countries.
William Wint (william.wint@zoo.ox.ac.uk) is an independent consultant and director of
the EnvironmentalResearch Group Oxford,which is based at the Department ofZoology,
University of Oxford, UK. He has worked extensively in the spatial analysisof many aspects of
the livestock sector, in particular livestockdistribution mapping and the modeling of disease
and vector distributions.
Stanley Wood (Stanley.Wood@gatesfoundation.org) is a senior program officer in the
Agricultural Development Program of the Bill & Melinda Gates Foundation, Seattle, Wash
ington. Earlier, he was a senior research fellow in the Environment and Production Tech
nology Division of IFPRI, where he led IFPRIs research on spatial analysis in a policy context
and served as coordinator of CGIARs Consortium for Spatial Information. Wood was also
one of the principal investigators for the HarvestChoice team.
Ulrike Wood-Sichra (u.wood-sichra@cgiar.org) has been a research analyst on the
HarvestChoice team at IFPRI, Washington, DC, since the teams inception. She deals with
crop-related data and models and the assembly of the HarvestChoice data matrix. Before
that, Wood-Sichra was a consultant on other projects at IFPRI and developed databases and
software. She also worked on agriculture and natural resources projects for earlier employers,
such as FAO, the Asian Development Bank, and the Inter-American Development Bank,
sometimes living in the countries the projects related to.
Sandra Yin (s.yin@cgiar.org) is an editor in the Communications and Knowledge
Management Division of IFPRI, Washington, DC.
Yisehac Yohannes (y.yohannes@cgiar.org) is a research analyst with the Poverty, Health,
and Nutrition Division of IFPRI, Washington, DC. He holds a master of science in food and
resource economics and a master of science in statistics. He has extensive experience in analyzing multifaceted large-scale primary household surveys that examine topics such as poverty and hunger alleviation, social protection, income, food consumption, and nutrition.
His current research involves analyzing agricultural growth and nutrition linkages and the
impacts of safety net programs.
Robert Zomer (R.Zomer@cgiar.org) is a landscape ecologist with a broad background
in plant community and forest and agricultural ecology and advanced skills in statistical analysis, geographic information systems, remote sensing, environmental modeling,
and landscape-level spatial analysis. He is currently a visiting professor at the Kunming
Institute of Botany, China, and a senior landscape ecologist at the World Agroforestry Centre
(ICRAF)-China.
About the Authors85
Glossary
agricultural systems: crop management schemes selected
by farmers to optimize the yield of a particular crop given
sociological, economic, biological, and political constraints.
agroecological zone: geographical areas that exhibit similar
climatic conditions that determine their ability to support
rainfed agriculture. These zones broadly define environments where specific agricultural systems thrive.
agropastoral farming systems: farming systems located
in semiarid areas of western, eastern, and southern Africa,
dominated by sorghum, millet, and livestock. Livelihoods are
derived from maize, pearl millet, pulses, sesame, sorghum,
cattle, goats, poultry, sheep, and off-farm activities.
aluminum toxicity: occurs in weathered soils that have
become highly acidic, making aluminum soluble and thus
toxic to plants. Aluminum toxicity is the most common soil
constraint in Africa south of the Sahara (SSA).
arable land: the land under temporary agricultural crops
(multiple-cropped areas are counted only once), temporary meadows for mowing or pasture, land under market
and kitchen gardens, and land temporarily fallow (less than
five years). This category does not include abandoned land
resulting from shifting cultivation.
arid: an area where the length of growing period (LGP) is less
than 70 days per year.
arid pastoral oasis farming systems: farming systems in
scattered communities in arid areas with average length of
growing period less than 30 days, and located primarily in
northwest, northeast, and southern Africa. Livelihoods
are based on cattle, small ruminants, date palms,
and off-farm activities.
aridity index: the ratio of annual total precipitation to
annual total potential evapotranspiration (PET). Aridity
index values increase with more humid conditions and
decrease with more arid conditions. The aridity index measures how much rainfall is available to satisfy the evapotranspiration water requirements for different reference
vegetation types.
blue water: water withdrawn from groundwater bodies (aquifers) or surface water bodies (rivers, lakes, wetlands, canals) and used for irrigation of agricultural land,
for drinking water, or by the industrial sector for processing
and cooling.
calcareous: a kind of soil that contains high levels of calcium carbonate. Calcareous soils can be highly fertile, but
extremely calcareous soils canlead to crop nutrient deficiencies by fixing phosphorus (see P fixation).
cereal-root crop mixed farming systems: farming systems
located in subhumid areas of western and central Africa,
distinguished by cereal crops along with roots and tubers.
Livelihoods are based on cassava, cattle, legumes, maize, millet, sorghum, yams, and off-farm activities.
coefficient of variation: a measure of variability from
an average calculated as the standard deviation divided
by the mean and expressed as a percentage, such as yearto-year rainfall variability.
Comprehensive Africa Agriculture Development
Program (CAADP): an Africa-led program designed to
promote increasing investments in agricultural growth in
Africa through research, extension, education, and training. CAADP is a program of the New Partnership for Africas
Development (NEPAD).
consumptive water use: in agriculture, typically refers to
crop evapotranspiration only and excludes return flows.
cracking clay:soils with high amounts of clay that shrink
and swell upon wetting and dryingalso called expansive
clay. These soils can be difficult to manage because they
can be too wet (reducing gas exchange in the soil) for good
plant growth. When wet, cracking clay can greatly expand
in volume and create additional soil problems. Extensive soil
cracking can disturb plant roots, and crusting can reduce
water infiltration, when dry.
crop evapotranspiration: the sum of evaporation from the
soil and transpiration of the plants.
dryland systems (also known as dryland agricultural
production systems): agroecosystems characterized by low
and erratic precipitation, persistent water scarcity, extreme
climatic variability, high susceptibility to land degradation
including desertificationand higher loss rates for natural resources, including biodiversity. In dryland systems, the
lack of water is the key factor that limits profitable agricultural production.
Ea/Et: ratio of actual to potential evapotranspiration.
Glossary87
evapotranspiration: the conversion of soil water into water

vapor. Estimating evapotranspiration rates is important
when planning irrigation schemes.
farming system: population of farm households that have
broadly similar resource and livelihood patterns, face similar
constraints and opportunities, and could benefit from similar development strategies and interventions. Household
livelihoods are based on farm production as well as offfarm activities.
fish-based farming systems: farming systems that are close
to major inland or coastal water bodies with fish as a major
source of livelihoods. Although located throughout Africa,
these fish-based farming systems are concentrated along
the coast and around major lakes. Livelihoods are based on
fish, bananas, cashews, coconuts, fruit, yams, poultry, goats,
forest-based farming systems: farming systems in humid
lowland, heavily forested areas of central Africa. Livelihoods
are based on subsistence food crops, including beans, cassava, cocoyams, maize, taro, and off-farm activities.
free of constraints: soils free from fertility constraints.
Global Hunger Index (GHI): a multidimensional measure
of hunger that combines three equally weighted indicators
(undernourishment, child underweight, and child mortality) in one index number. It takes into account the nutrition
situation of not only the population as a whole, but also of
a physiologically vulnerable groupchildrenfor whom a
lack of nutrients creates a high risk of illness, poor physical
and cognitive development, and/or death.
Global Yield Gap Atlas Extrapolation Domain
(GYGA-ED): a climate zone scheme or domain based on
three variables: (1) growing degree days with base temperature of 0C; (2) temperature seasonality (quantified as
the standard deviation of monthly average temperatures);
and (3) an aridity index (annual total precipitation divided
by annual total potential evapotranspiration). (See aridity
index, potential evapotranspiration, and transpiration).
green water: precipitation stored in the soil and used by
rainfed and irrigated crops.
growing degree days (GDD): a measure of heat accumulationused to estimateplant development rates.GDD are
calculated as the difference between current temperatures and a minimum base threshold temperature (where
growth rate=0). Plant growth rates can be measured through
88
the accumulation of GDD, with different species requiring

different numbers of accumulated GDD to reach maturity.
highland mixed farming systems: farming systems in cool
highland areas (above 1,600 meters), dominated by temperate cereals and livestock, located in eastern and southern
Africa. Livelihoods are based on broadbeans, goats, lentils,
peas, potatoes, rape, teff, wheat barley, poultry, sheep, and
off-farm activities.
highland perennial farming systems: farming systems in
moist highland areas (above 1,400 meters) of eastern Africa,
with relatively good market access and with a dominant
perennial crop, either food or commercial. Livelihoods are
based on diverse activities, including bananas, beans, cassava, coffee, enset (or false banana, Enset ventricosum, in
Ethiopia), maize, sweet potatoes, tea, livestock (including
dairy), and off-farm activities.
humid: an area where the length of growing period (LGP) is
greater than 270 days per year.
humid lowland tree crop farming systems: farming systems located in western and central Africa that appear in
humid lowland areas where commercial tree crops have
replaced forest and provide more than one-quarter of
household cash income. Livelihoods are based on coffee,
cocoa, oil palm, and rubber, as well as cassava, maize, yams,
insurance crops: crops that increase food security because
they can be left in the ground until needed. Roots and
tubers, including cassava, fall into this category.
intensity ratio of investment: public R&D investment measured as a share of agricultural output.
irrigated farming systems: large-scale contiguous irrigation
schemes, with almost no rain-fed agriculture. Located mostly
in areas with low rainfall. Livelihoods are largely based on
irrigated commercial crops, notably rice, cotton, and vegetables, as well as cattle and small ruminants.
land cover: the physical material at the surface of the
earth, such as crops, pasture, trees, bare rock, water, and
urban areas.
leaching:occurs when water percolating through the soil
moves soluble nutrients below the crop root zones. Over
time leaching can reduce the availability of nutrients to
crops. Old, highly weathered soils in areas of moderate to
high precipitation are typically nutrient depleted and acidic
as a result of nutrient leaching.
length of growing period (LGP): generally calculated as the

period (in days) during a year when precipitation exceeds
half the potential evapotranspiration, while also taking into
account soil moisture holding capacity. It is used to determine the number of days per year that are suitable for crop
growth in a given location.
livestock system: a farming system where more than 90 percent of dry matter fed to animals comes from rangelands,
pastures, annual forages, and purchased feed and less than
10 percent of the total value of production comes from nonlivestock farming activities.
low nutrient reserves: soils with less than 10 percent
reserves of weatherable minerals that naturally supply phosphorus, potassium, calcium, magnesium and micronutrients.
major climate divisions: major latitudinal thermal or temperature shifts in climate zones.
maize mixed farming systems: farming systems located in
subhumid and humid areas of eastern, middle, and southern
Africa, dominated by maize with legumes. Livelihoods are
based mainly on maize, tobacco, cotton, legumes, cassava,
cattle, goats, poultry, and off-farm activities.
marketshed: geographical area and associated population
that has real or potential trade relationships with a market
center. Each market shed is associated with the closest corresponding market in terms of the least-cost travel time to
that market.
MarkSim: a statistical weather generator that produces
weather records (rainfall, maximum and minimum air temperature, and solar radiation) on a daily basis. It is able to
simulate the variation in rainfall observed in both tropical
and temperate regions.
New Partnership for Africas Development (NEPAD):

a vision and a policy framework of the African Union
for pan-African socioeconomic development in the
21st century.
North Africa dryland mixed farming systems: farming systems in dry semi-arid areas with rainfall of 150300 mm,
based on rainfed barley and wheat grown in a rotation
with one- or two-year fallows and a strong small ruminant
component. Livelihoods also include off-farm activities.
North Africa highland mixed farming systems: farming
systems dominated by rainfed cereal and legume cropping
with tree crops, fruits, and olives on terraces, together with
vines and/or raising livestock (mostly sheep) on communally
managed lands and characterized by moderately high population densities. Livelihoods also include off-farm activities.
North Africa rainfed mixed farming systems: farming systems in subhumid areas characterized by tree
crops (olive and fruit), melons, grapes, irrigated vegetables, and flowers as well as rainfed wheat, barley, chickpea, lentil, and fodder crops. Livelihoods are supplemented
by dry-season grazing of sheep migrating from the steppe
areas and off-farm activities.
P fixation: occurs when phosphorus (P) becomes insoluble
and therefore is not available to plants. Extremely calcareous
soils, which contain high levels of calcium carbonate, and
soils that are rich in iron and aluminum oxides fix phosphorus and canlead to nutrient deficiencies in a crop.
pastoral farming systems: farming systems with low
population density in arid areas of western, eastern, and
southern Africa, dominated by livestock. Livelihoods are
based on camels, cattle, goats, sheep, some cereal crops,
mixed crop-livestock farming systems: a farming system in which more than 10 percent of the dry matter fed to
animals comes from crop by-products (for example, stubble) or more than 10 percent of the total value of production comes from nonlivestock farming activities. Livestock
convert organic material not fit for human consumption
into high-value food products (meat, milk) and nonfood
products (traction, manure, leather, bone).
perennial mixed farming systems: commercially oriented

farming systems predominantly found in South Africa and
comprising deciduous fruits and vineyards in the Western
Cape and eucalyptus, pines, and wattle as well as sugarcane
in the southeastern region (KwaZulu-Natal, Mpumalanga and
Eastern Cape Provinces) interspersed with cereals, oilseeds,
and pulses. Livelihoods include off-farm activities.
net primary production (NPP): the amount of biomass

produced by a plant or ecosystem, excluding the energy
it uses for the process of respiration. This typically corresponds to the rate of photosynthesis, minus respiration by
the photosynthesizers.
permanent crops: cropssuch as cocoa, coffee, and rubberthat are sown or planted once and then occupy the
land for several years and do not need to be replanted after
each annual harvest. This category includes flowering shrubs,
fruit trees, nut trees, and vines, but excludes trees grown for
wood or timber.
Glossary89
permanent meadows and pastures: land used five years or

more to grow herbaceous forage crops, either cultivated or
growing wild (wild prairie or grazing land).
subhumid: an area where the length of growing period

(LGP) is 180270 days per year.
poor drainage: soils characterized by the inability to properly drain.
subtropics: areas where mean monthly temperature

adjusted to sea-level is less than 18 C for one or more
months in a year.
portshed: an area associated with the closest corresponding

port in terms of the least-cost travel time to that port.
transpiration: the evaporation of water from the leaves and

stems of plants.
potential evapotranspiration (PET): the energy available in

the system to remove water through the processes of evaporation and transpiration. It is generally associated with a
reference crop, namely short grass completely covering the
ground, and assumes no limitation on water availability.
tropics: areas where the monthly temperature adjusted

to sea-level is greater than 18 C for all months.
rain-use efficiency (RUE): the amount of biomass produced (kilograms of dry matter per hectare) per millimeter
of rainfall calculated as the ratio of net primary production
(NPP) over rainfall.
root and tuber crop farming systems: farming systems
located in lowland areas of western and middle Africa where
systems are dominated by roots and tubers without a major
tree crop. Livelihoods are based mainly on cassava, legumes,
yams, and off-farm activities.
seasonality: the way in which climate (such as rainfall or
temperature) varies regularly through the year in a particular place.
semiarid: an area where the length of growing period (LGP)
is 70180 days per year.
Spatial Production Allocation Model (SPAM): a model
that produces estimates of crop distribution and can be used
to generate maps showing area harvested per cell by crop
and production system (technology). The model draws on
many datasets, including land cover imagery, crop suitability
maps, irrigation maps, subnational crop statistics, FAO country totals of crop production and area, and data on production systems in each country.
stem rust: a fungal disease that affects wheat.
stunting: low height for age in children (under age five).
Stunting reflects a sustained past episode or episodes of
chronic undernutrition.
90
trypanosomosis: a parasitic disease transmitted by the tsetse fly. The African animal form of the disease reduces the
productivity of livestock, especially cattle, when it sickens or
kills them.
Ug99: the collective name for new strains of stem rust
pathogen, first discovered in Uganda in 1998. Most of the
worlds wheat varieties offer little resistance to Ug99 (see
stem rust).
undernutrition: a measure of food energy deprivation.
Undernutrition results when prolonged food energy intake is
below standard nutritional requirements and/or low levels of
absorption of food consumed.
underweight: low weight for age in children (under age
five). Underweight reflects a current condition resulting
from inadequate food intake, past episodes of undernutrition, and/or poor health conditions.
virtual water: the water needed to produce a product. If
a country exports such a product, it exports water in virtual form.
virtual water content: the volume of water used by a crop
per unit of crop harvest.
volcanic: amorphous soils characterized by large reserves
of weatherable minerals (which are unstable in humid climates) and soil organic matter making them highly fertile.
Volcanic soils also have a high phosphorus fixation capacity
which can slightly limit their fertility.
wasting: low weight for height in children under age five.
Wasting generally reflects an acute weight loss associated
with a recent period of hunger or disease.
ATLAS
OF AFRICAN
AGRICULTURE RESEARCH
& DEVELOPMENT
T
he work of agricultural researchers and development

workers in Africa has the potential to significantly
improve the lives of the poor. But that potential can only
be realized with easy access to high-quality data and
information. The Atlas of African Agriculture Research &
Development highlights the ubiquitous role of smallholder
agriculture in Africa; the many factors shaping the location,
nature, and performance of agricultural enterprises; and
the strong interdependencies among farming, naturalresource stocks and flows, and the well-being of the poor.
Organized around 7 themes, the atlas covers more
than 30 topics, each providing mapped geospatial data
and supporting text that answers four fundamental
IFPRI
INTERNATIONAL
FOOD POLICY
RESEARCH
INSTITUTE
2033 K Street, NW, Washington, DC 20006-1002 USA

Tel.: +1.202.862.5600 SKYPE: ifprihomeoffice
Fax: +1.202.467.4439 Email: ifpri@cgiar.org
www.ifpri.org
questions: What is this map telling us? Why is this

important? What about the underlying data? Where can
I learn more?
The atlas is part of a wide-ranging eAtlas initiative
that will showcase, through print and online resources, a
variety of spatial data and tools generated and maintained
by a community of research scientists, development
analysts, and practitioners working in and for Africa. The
initiative will serve as a guide, with references and links
to online resources to introduce readers to a wealth of
data that can inform efforts to improve the livelihoods
of Africas rural poor. To learn more about the eAtlas
initiative, visit http://agatlas.org.
CGIARCSI
Consortium for Spatial Information
R180
G46
B52
R154
G58
B32
R255
G204
B104
R52
G51
B22
R227
G137
B27
R198
G113
B41

Atlasafricanag All 1

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Atlasafricanag All 1

Uploaded by

Copyright:

Available Formats

ATLAS

ABOUT THE MAPS

Library of Congress Cataloging-in-Publication Data

COVER PHOTO CREDITS

About the Authors 81

Abbreviations and Acronyms

master of science degree

OF AFRICAN AGRICULTURE RESEARCH & DEVELOPMENT

POLITICAL, DEMOGRAPHIC, AND

POLITICAL, DEMOGRAPHIC, AND INSTITUTIONAL CLASSIFICATIONS

WHY IS THIS IMPORTANT?

FIGURE 1 Accuracy of different administrative boundary

Number of countries with

WHAT IS THIS MAP TELLING US?

Source: Adapted from Brigham, Gilbert, and Xu 2013.

Knowing that the same boundaries were used across the

WHAT ABOUT THE UNDERLYING DATA?

WHERE CAN I LEARN MORE?

OF AFRICAN AGRICULTURE RESEARCH & DEVELOPMENT

Country and first-level administrative boundaries

Data source: FAO 2013.

POLITICAL, DEMOGRAPHIC, AND INSTITUTIONAL CLASSIFICATIONS

Stanley Wood and Kate Sebastian

WHY IS THIS IMPORTANT?

Cereal yield (tons per hectare)

The agriculture research and development community

FIGURE 1 Comparing different aggregates,

WHAT ARE THESE MAPS TELLING US?

Data source: FAO 2012a; FAO 2012b; World Bank 2013b.

WHAT ABOUT THE UNDERLYING DATA?

WHERE CAN I LEARN MORE?

OF AFRICAN AGRICULTURE RESEARCH & DEVELOPMENT

FAO regional groups

World Bank income groups

POLITICAL, DEMOGRAPHIC, AND INSTITUTIONAL CLASSIFICATIONS

Gert-Jan Stads, Nienke Beintema, and Kathleen Flaherty

WHY IS THIS IMPORTANT?

FIGURE 1 Donor funding as a share of total agriculture

WHAT ABOUT THE UNDERLYING DATA?

WHERE CAN I LEARN MORE?

OF AFRICAN AGRICULTURE RESEARCH & DEVELOPMENT

No data or non-ASTI country

Data source: ASTI 2013.

No data or non-ASTI country

MAP 2Intensity of agriculture R&D spending, 2011

MAP 1 Change in public agriculture R&D spending levels,

POLITICAL, DEMOGRAPHIC, AND INSTITUTIONAL CLASSIFICATIONS

Nienke Beintema, Gert-Jan Stads, and Kathleen Flaherty

WHY IS THIS IMPORTANT?

FIGURE 1 Age and sex structure of agricultural R&D

WHAT ARE THESE MAPS TELLING US?

Source: Sne et al. 2012.

WHAT ABOUT THE UNDERLYING DATA?

WHERE CAN I LEARN MORE?

OF AFRICAN AGRICULTURE RESEARCH & DEVELOPMENT

No data or non-ASTI country

Total FTE researchers

MAP 2 Share of agricultural researchers with

MAP 1Number of agricultural researchers, 2011

MAP 3Concentration of agricultural researchers, 2011

No data or non-ASTI country

POLITICAL, DEMOGRAPHIC, AND INSTITUTIONAL CLASSIFICATIONS

WHAT IS THIS MAP TELLING US?

About the Authors 81

MAP 2Intensity of agriculture R&D spending, 2011

MAP 1Number of agricultural researchers, 2011

MAP 3Concentration of agricultural researchers, 2011